#or maybe it will be a cycle of ai makes weird meme > memes about the ai meme pop up > ai learns from said memes > so on

[dexaroth]

I wonder how badly AI will impact the storage spaces of basically every website given how um. Immediate you can make them images

guess we should start bracing for mass inactive account deletions (harder than before ofc bc its an inevitability)

[meta-squash]

Lipstick Traces Review/Thoughts

(I wrote this 2 years ago but didn’t have a tumblr to post it to at the time)

So I’ve just finished reading Lipstick Traces by Greil Marcus. And it’s fucking long with so much information and I’ve been having a lot of thoughts. Some just about little specific things mentioned in the book, and some more about the themes of the book written in the 80s compared to our current epoch of technology and politics and art and culture industry etc.

I mean, a lot of the stuff in the book/the thoughts the book gave me are things I’ve rambled about before on tumblr. But I guess it’s stuff that’s still in my head, that still bothers me, that I still have no solution for, or that I can find cracks in my arguments for solutions.

Mostly what I took away from this book was a giant feeling of conflict and ambivalence and uncertainty. It is, ultimately, a book of regret. It’s a book that explores these artists and movements and ideas and people that made a series of tiny but huge impacts to art and creation, who could have made a huge revolutionary change, but whose small revolutions were lost to time. It is a book about anger or frustration that incites a change, an avant garde, and how that anger fizzles out or is smothered and forgotten. It is a book about the cycles of history and how the new, the angry, the ones pushing back, are always eventually suppressed. In a 1994 quote Richey said, essentially, that you only really get remembered if you’re an Einstein or a Newton– a person who creates or discovers something that is such a massive revolutionary change that it affects the way the world is perceived and how it is believed to function. This book talks about those who aren’t Newtons and Einsteins. Those artists that made little waves that changed a few but didn’t change enough.

And it’s simultaneously fascinating and exciting and depressing, reading and thinking about this. That this book is a book of regret written in the 1980s, and 35 years later things have only gotten more extreme, and the regret can only feel heavier. The anger is still there, too, but it was more potent in the 80s and 90s, it had more potential. Now the anger is becoming impotent, or trapped. Either the meek inherited the earth and forgot what it was like to be meek, or the ones who inherited the earth were strongmen wearing the masks of the meek and the ambition of the avant garde.

Honestly, the biggest takeaway I got from this book is how drastically things have changed. How the way the book compares the Dadaists to the original punks is a fairly close, similar type of comparison, with similar movements, ideas, ideals, messages, and actions. And how the comparison to both of those with any sort of movement that might happen in the next decade or so will be massively, drastically different because of how much culture has changed, media has changed, access and accessibility has changed, government, education, class awareness, and on and on. How, honestly, I’m not sure if there could be another movement like the dadaists and like the punk scene, because to be reactionary and avant garde and revolutionary is something very different these days.

Already Greil Marcus discusses speed and the culture industry. Which makes sense, since his primary theoretical sources are Guy Debord and Theodor Adorno. But it’s fascinating to see these theories–both written and published in the 40s and 60s–being used to critique and analyse culture and art back then, much closer to the texts’ inception. Those theories were new-ish in terms of being put into words back then. The idea of the prison of capitalism, the labor that turns the proletariat into machines and then sells them back to themselves, the speed and change of media, the homogenous nature of entertainment and pop culture. All of that was relatively new, at least in terms of being stated outright.

And people were frustrated! People have always been frustrated! The Dadaists were frustrated by the war they didn’t want to participate in, and then in the monotony of the post-war expectations that everything go back to normal, when nothing was normal. They were frustrated by the Modernists, by the Expressionists, by art becoming something that gave you Status rather than something that you just did because you had the urge. Punks were frustrated with the economic and social malaise, the labor issues, the failed ideals of the hippies, art and music stagnating, the lack of platforms for them to express themselves. But they were able to use art to express that anger, that frustration, that feeling of nihilism or of glee at meaninglessness, that feeling of “fuck it, we have nothing so let’s do what we want.” Both generations did it in different styles, but both threw convention out the window, focused on what was taboo, what was weird, what was scandalous, what they wanted to say but society didn’t want them saying.

What’s interesting about the book is that it expresses admiration for this, for the daring and avant garde and original and clever and badass nature of both Dada and Punk ideals/styles/philosophies/actions/etc. But it also expresses regret. Regret that it only lasted so long. That it didn’t leave any major effect on art or politics or life or society (that is, aside from what capitalism stole or what minor underground movements admired or were inspired by). That it was stolen by capitalism. That it inevitably fell apart as time moved forward.

But for those glorious few years….

And what it made me think of, which (like I said) Marcus talks about quite a bit, is the effect that the culture industry and the speed of culture/media/news had on both movements. For the Dadaists, it was more about the speed of the news and also just blindly making, with no knowledge of a goal or ultimate desire, that resulted in the group eventually separating into other factions and the movement petering out into other artistic ideas and styles. The Dadaists were reacting to the war, to the newness of certain parts of culture, to the personal conflicts between artists. The punk movement was more affected by the ever-increasing speed of culture and media as well as news. Things were moving faster. Styles and ideas were coming into fashion and then becoming old hat more quickly. Punk started out as avant-garde, as a refusal to conform, as an excuse and/or reason to speak out and act out and express oneself. Especially in communities that weren’t being heard. It started out as a way for individuals to force society to acknowledge them. And then capitalism and the culture industry got their hands on it and began to use it as a marketing ploy, as fashion, selling punk back to the masses it was intended to belong to.

It’s pretty obvious that the world has sped up immensely since the 1970s– media, news, and culture industry included. Things that are new on Monday are old by Friday. Memes that are hilarious and circulate social media for weeks are dead by the time companies try to capitalize on them (see: Zumiez etc making Grumpy Cat shirts etc). Music or films that are popular fall out of popularity in just a few weeks, unless they’re vapid pieces of media or unless the creators/artists continue to hype themselves over and over again in different ways. It is impossible to create focused critical art because there is always so much going on in the news and in world politics or social issues; everything is so intertwined it’s impossible to pick out certain things to criticize. Artists and art movements and things of meaning and import fall by the wayside. It’s hard for me to imagine an avant garde or artistic movement within a community growing in popularity and staying strong for long enough to really make an impact or a difference. And the speed of the news is insane now. Things are only big news for a few days before vanishing under the avalanche of new stories and new events. Things stay big news within the communities that care about them (ie Black Lives Matter, Flint MI, Grenfell, DAPL, etc) but not within the eye of the media. News changes as fast as a feed can refresh.

I also have the feeling that art doesn’t have as much power. Subliminal marketing power, sure. But the last few art pieces I remember hearing even random people talking about were Shepard Fairey’s 2008 portraits for the Obama campaign, Ai Weiwei’s Han dynasty vase smash (which was from 1995 but came back into the spotlight in the mid-2000s for some reason) and Yayoi Kusama’s infinity mirrored room. It’s hard now, with the constant barrage of information and images and sounds, to figure out what is important and impactful art, and what is rubbish (or advertisement). It’s also hard to figure out what to focus on when making critical art: what moments or events in politics and current events will be remembered long enough to be used in critique; what will people remember and be affected by? Maybe hindsight is 20/20 tunnel vision or the gaze towards the past is tinged with roses, but it seems as though art had a larger significance. Barbara Kruger, for example. The Sex Pistols, The Guerilla Girls, Robert Mapplethorpe, Keith Haring, Annie Liebowitz, and (obviously) Jenny Holzer. All used their art to critique various current events, social/political/global issues. They had an effect on viewers in their time as well as after it. It seems as though, now, there’s no during-and-after. There is only during (like Shepard Fairey’s portraits).

A big reason for that, I think, is because of the disintegration of Dadaism and Situationism due to speed and capitalism. Basically, Situationism was created to force those going about their daily lives to stop for a second and think about their situation, to make a moment of “real living,” to jolt people out of the stupor of the daily grind and make them remember. Remember they’re alive, remember they shouldn’t be living a life of a drone, remember they’re consuming things they’re being told to instead of doing what they want to. And those moments were created through graffiti, through the detournement of taking normal comic strips and rewriting dialogues to critique the world, through the music and fashion of punk, which shouted out the flaws in society without caring that it was supposed to be kept hush-hush, through visual art that confronted the viewer with critiques (like Barbara Kruger or Jenny Holzer), etc etc. But now, do something like that and you’re called “edgy” and mocked. Why? Probably because of the likes of Banksy. I say this because Banksy often creates graffiti pieces that probably should or would have meaning, or should or would make you stop and think. Except that they’re pieces by Banksy, famous for being edgy, whose pieces are worth thousands or millions of dollars. Who rarely actually has a statement, except money-making. How many of us howled with laughter when he made that nightmare-Disneyland piece? Because it was edgy and unoriginal. Because we already know we’re living in a slowly growing dystopia, and being told that by a guy who benefits from said dystopia and gets so much money from criticizing it is bullshit.

It’s also because it feels like there’s nothing new under the sun. Now, Greil Marcus kind of talks about this. The punk movement expressed this too. The nihilism that nothing is new, that everything has already been said. But it did so gleefully, embracing the nihilism in order to laugh at it and point it out and roll in that glee. There is nothing new to be said, they thought, but there are new ways to say it. Because we’ve been saying things for centuries but nothing has changed, except the way it gets said. The problem now, in the 21st century, is that nothing new under the sun is now nothing new under the sun and that can no longer be used as a statement. “It’s all already been done, just say it in a new way” is no longer good enough. Ideas have to come out of a vacuum— except if they come out of a vacuum, they’re either never noticed or they’re appropriated by the media and capitalism.

Basic Adorno, basic culture industry theory. But Adorno would have a fucking aneurysm if he could see how his theory holds up in the 21st century compared to 1944. And honestly, that is a terrifying sentence to type. That Adorno and Horkheimer published Enlightenment as Mass Deception in 1944, that they were noticing this in the 1940s. And every point in their essay has only increased exponentially since then.

Greil Marcus hints at the whole “punk is dead” thing throughout the book without actually saying those words. I don’t think the phrase really existed as a buzzword type thing when the book was published. But I think the points and ideas expressed in Lipstick Traces kind of say what my thoughts have always been on that idea. Punk is dead, and punk is also not dead. Punk is dead; its looks and sound were stolen by the media and by capitalism and sold to the masses, sold back to the kids who created and popularized it. Punk was the sound and creativity and style of the kids who had nothing and wanted to be everything, so they made it all themselves. They created their own style and said what they wanted to say. High fashion stole it, television stole it, department stores stole it, ad agencies stole it, and sold it back. “Ever get the feeling you’ve been cheated?” Punk is dead, as an original movement, as an original fashion. But! But, punk thought is not. Punk as an ideal, as a philosophy, as thought, is very much alive. Punk, as the idea that you make your own, that you use your own creativity and express yourself the way you want to. That it’s passion and not necessarily talent that matters. That wearing what you want, saying what you want, confronting the issues that need confronting, being whoever you are so long as you’re not hurting or fucking over an innocent person, that’s still very much alive. The original punk fashion has been stolen. But punk fashion still exists, in people that make their own clothes or wear strange things even though they get stared at. Punk in art still exists, in people that make their art for themselves, or who make art with friends despite knowing they might go nowhere, just because they have the passion. Punk music is the same. The ideals and thought is still thrumming and alive. Its parent has been consumed by consumerism, devoured by capitalism and marketing and fashion. But the orphaned offspring is still hiding and alive.

And yet there’s another ‘but.’ The depressing one. Which is that it feels as though punk, in the early, original days, gave the youth a label, an identity. This goes for plenty  of other youth movements as well, and art movements, etc etc. But these days it seems a community identity hardly exists. And it’s hard to push a movement, create a feeling of community or solidarity, without some sort of shared identity. Perhaps the label of “Millenials” and “Gen Z” are the closest we’ve come so far. But those are so broad, and so often used in a derogatory fashion (although, I suppose, so were “punk” and “mod” and “hippie” and “teddy boy” etc etc).

And I also think that everything is so fast now, and moments are so fleeting, events are so quick to be forgotten, that it is hard to impress an idea or affect change or put an artistic statement or movement out there for long enough to make a true impact. I would say that maybe a large amount of the generation(s) banding together to make a statement would do something, would make that change. But Black Lives Matter was made up mostly of Millenials, young people, people under the age of 35. And yet it slowly petered away into almost nothingness with no changes.

But the kids of the next generation, Gen Z, do give me hope. Like that other person’s post going around says, they’re pissed, they were raised on a steady diet of dystopian literature with strong main characters, they’re highly aware of the state of politics and the job market and the economy, they’ve seen how fucked Millenials are and they know it’s not going to get much better for a while. And maybe they’re the next ones, the next to say “fuck it, we have nothing and we are nothing, let’s do whatever we want because we haven’t got anything to lose”. And maybe the millenials will join.

That’s what I hope. That’s what Greil Marcus’ book seems to be trying to say. That these sorts of movements don’t always have massive, lasting effects in the grand scheme of the world and society. But they leave cracks, and fragments, and shrapnel, and artefacts, for the next generation or the next movement to find and use. That dadaism might have faded away and punk might be dead but the dadaist yell is still echoing and punk thought is still very much alive. And it’s up to us to hear it, to use it, to find the crack in the culture industry and capitalism and society and somehow find the next avant garde, the ideas and movement that will stick and create an identity for unfettered expression, if only for a little while. That “the moment of real poetry brings all the unsettled debts of history back into play,” and it is up to us to figure out what we have to do or say to ignite all of that history and to wield its power. And how we can make our own history or try and settle the debts of the past.

(And yet…. And yet…. And yet I can’t help but doubt that the speed of the world will allow this to happen. And yet I want to believe that something can be done to create critical work that sticks. And yet how do you make critical work without it being eaten up by the culture industry and disappeared into homogeneity. And yet we have technology and creative mediums now that we didn’t in 1977. And yet punk is dead. And yet punk thought is not. And yet, and yet.)

[smartwebhostingblog]

Autocomplete Presents the Best Version of You

New Post has been published on http://webhostingtop3.com/autocomplete-presents-the-best-version-of-you/

Autocomplete Presents the Best Version of You

Type the phrase “In 2019, I’ll …” and let your smart­phone’s keyboard predict the rest. Depending on what else you’ve typed recently, you might end up with a result like one of these:

In 2019, I’ll let it be a surprise to be honest. In 2019, i’ll be alone. In 2019, I’ll be in the memes of the moment. In 2019, I’ll have to go to get the dog. In 2019 I will rule over the seven kingdoms or my name is not Aegon Targareon [sic].

Many variants on the predictive text meme—which works for both Android and iOS—can be found on social media. Not interested in predicting your 2019? Try writing your villain origin story by following your phone’s suggestions after typing “Foolish heroes! My true plan is …” Test the strength of your personal brand with “You should follow me on Twitter because …” Or launch your political career with “I am running for president with my running mate, @[3rd Twitter Suggestion], because we …”

Gretchen McCulloch is WIRED’s resident linguist. She’s the cocreator of Lingthusiasm, a podcast that’s enthusiastic about linguistics, and her book Because Internet: Understanding the New Rules of Language is coming out in July 2019 from Penguin.

In eight years, we’ve gone from Damn You Autocorrect to treating the strip of three predicted words as a sort of wacky but charming oracle. But when we try to practice divination by algorithm, we’re doing something more than killing a few minutes—we’re exploring the limits of what our devices can and cannot do.

Your phone’s keyboard comes with a basic list of words and sequences of words. That’s what powers the basic language features: autocorrect, where a sequence like “rhe” changes to “the” after you type it, and the suggestion strip just above the letters, which contains both completions (if you type “keyb” it might suggest “keyboard”) and next-word predictions (if you type “predictive” it might suggest “text,” “value,” and “analytics”). It’s this predictions feature that we use to generate amusing and slightly nonsensical strings of text—a function that goes beyond its intended purpose of supplying us with a word or two before we go back to tapping them out letter by letter.

The basic reason we get different results is that, as you use your phone, words or sequences of words that you type get added to your personal word list. “For most users, the on-device dictionary ends up containing local place-names, songs they like, and so on,” says Daan van Esch, a technical program manager of Gboard, Google’s keyboard for Android. Or, in the case of the “Aegon Targareon” example, slightly misspelled Game of Thrones characters.

Another factor that helps us get unique results is a slight bias toward predicting less frequent words. “Suggesting a very common word like ‘and’ might be less helpful because it’s short and easy to type,” van Esch says. “So maybe showing a longer word is actually more useful, even if it’s less frequent.” Of course, a longer word is probably going to be more interesting as meme fodder.

Finally, phones seem to choose different paths from the very beginning. Why are some people getting “I’ll be” while others get “I’ll have” or “I’ll let”? That part is probably not very exciting: The default Android keyboard presumably has slightly different predictions than the default iPhone keyboard, and third-party apps would also have slightly different predictions.

Whatever their provenance, the random juxtaposition of predictive text memes has become fodder for a growing genre of AI humor. Botnik Studios writes goofy songs using souped-up predictive keyboards and a lot of human tweaking. The blog AI Weirdness trains neural nets to do all sorts of ridiculous tasks, such as deciding whether a string of words is more likely to be a name from My Little Pony or a metal band. Darth Vader? 19 percent metal, 81 percent pony. Leia Organa? 96 percent metal, 4 percent pony. (I’m suddenly interpreting Star Wars in quite a new light.)

The combination of the customization and the randomness of the predictive text meme is compelling the way a BuzzFeed quiz or a horoscope is compelling—it gives you a tiny amount of insight into yourself to share, but not so much that you’re baring your soul. It’s also hard to get a truly terrible answer. In both cases, that’s by design.

You know how when you get a new phone and you have to teach it that, no, you aren’t trying to type “duck” and “ducking” all the time? Your keyboard deliberately errs on the conservative side. There are certain words that it just won’t try to complete, even if you get really close. After all, it’s better to accidentally send the word “public” when you meant “pubic” than the other way around.

This goes for sequences of words as well. Just because a sequence is common doesn’t mean it’s a good idea to predict it. “For a while, when you typed ‘I’m going to my Grandma’s,’ GBoard would actually suggest ‘funeral,'” van Esch says. “It’s not wrong, per se. Maybe this is more common than ‘my Grandma’s rave party.’ But at the same time, it’s not something that you want to be reminded about. So it’s better to be a bit careful.”

Users seem to prefer this discretion. Keyboards get roundly criticized when a sexual, morbid, or otherwise disturbing phrase does get predicted. It’s likely that a lot more filtering happens behind the scenes before we even notice it. Janelle Shane, the creator of AI Weirdness, experiences lapses in machine judgment all the time. “Whenever I produce an AI experiment, I’m definitely filtering out offensive content, even when the training data is as innocuous as My Little Pony names. There’s no text-generating algorithm I would trust not to be offensive at some point.”

The true goal of text prediction can’t be as simple as anticipating what a user might want to type. After all, people often type things about sex or death—according to Google Ngrams, “job” is the most common noun after “blow,” and “bucket” is very common after “kick the.” But I experimentally typed these and similar taboo-but-common phrases into my phone’s keyboard, and it never predicted them straightaway. It waited until I’d typed most of the letters of the final word, until I’d definitely committed to the taboo, rather than reminding me of weighty topics when I wasn’t necessarily already thinking about them. With innocuous idioms (like “raining cats and”), the keyboard seemed more proactive about predicting them.

Instead, the goal of text prediction must be to anticipate what the user might want the machine to think they might want to type. For mundane topics, these two goals might seem identical, but their difference shows up as soon as a hint of controversy enters the picture. Predictive text needs to project an aspirational version of a user’s thoughts, a version that avoids subjects like sex and death even though these might be the most important topics to human existence—quite literally the way we enter and leave the world.

We prefer the keyboard to balance raw statistics against our feelings. Sex Death Phone Keyboard is a pretty good name for my future metal band (and a very bad name for my future pony), but I can’t say I’d actually buy a phone that reminds me of my own mortality when I’m composing a grocery list or suggests innuendos when I’m replying to a work email.

The predictive text meme is comforting in a social media world that often leaps from one dismal news cycle to the next. The customizations make us feel seen. The random quirks give our pattern-seeking brains delightful connections. The parts that don’t make sense reassure us of human superiority—the machines can’t be taking over yet if they can’t even write me a decent horoscope! And the topic boundaries prevent the meme from reminding us of our human frailty. The result is a version of ourselves through the verbal equivalent of an Instagram filter, eminently shareable on social media.

More Great WIRED Stories

Tech

[awesomecolorfulcollectorwombat]

Autocomplete Presents the Best Version of You

New Post has been published on http://webhostingtop3.com/autocomplete-presents-the-best-version-of-you/

Autocomplete Presents the Best Version of You

Type the phrase “In 2019, I’ll …” and let your smart­phone’s keyboard predict the rest. Depending on what else you’ve typed recently, you might end up with a result like one of these:

In 2019, I’ll let it be a surprise to be honest. In 2019, i’ll be alone. In 2019, I’ll be in the memes of the moment. In 2019, I’ll have to go to get the dog. In 2019 I will rule over the seven kingdoms or my name is not Aegon Targareon [sic].

Many variants on the predictive text meme—which works for both Android and iOS—can be found on social media. Not interested in predicting your 2019? Try writing your villain origin story by following your phone’s suggestions after typing “Foolish heroes! My true plan is …” Test the strength of your personal brand with “You should follow me on Twitter because …” Or launch your political career with “I am running for president with my running mate, @[3rd Twitter Suggestion], because we …”

Gretchen McCulloch is WIRED’s resident linguist. She’s the cocreator of Lingthusiasm, a podcast that’s enthusiastic about linguistics, and her book Because Internet: Understanding the New Rules of Language is coming out in July 2019 from Penguin.

In eight years, we’ve gone from Damn You Autocorrect to treating the strip of three predicted words as a sort of wacky but charming oracle. But when we try to practice divination by algorithm, we’re doing something more than killing a few minutes—we’re exploring the limits of what our devices can and cannot do.

Your phone’s keyboard comes with a basic list of words and sequences of words. That’s what powers the basic language features: autocorrect, where a sequence like “rhe” changes to “the” after you type it, and the suggestion strip just above the letters, which contains both completions (if you type “keyb” it might suggest “keyboard”) and next-word predictions (if you type “predictive” it might suggest “text,” “value,” and “analytics”). It’s this predictions feature that we use to generate amusing and slightly nonsensical strings of text—a function that goes beyond its intended purpose of supplying us with a word or two before we go back to tapping them out letter by letter.

The basic reason we get different results is that, as you use your phone, words or sequences of words that you type get added to your personal word list. “For most users, the on-device dictionary ends up containing local place-names, songs they like, and so on,” says Daan van Esch, a technical program manager of Gboard, Google’s keyboard for Android. Or, in the case of the “Aegon Targareon” example, slightly misspelled Game of Thrones characters.

Another factor that helps us get unique results is a slight bias toward predicting less frequent words. “Suggesting a very common word like ‘and’ might be less helpful because it’s short and easy to type,” van Esch says. “So maybe showing a longer word is actually more useful, even if it’s less frequent.” Of course, a longer word is probably going to be more interesting as meme fodder.

Finally, phones seem to choose different paths from the very beginning. Why are some people getting “I’ll be” while others get “I’ll have” or “I’ll let”? That part is probably not very exciting: The default Android keyboard presumably has slightly different predictions than the default iPhone keyboard, and third-party apps would also have slightly different predictions.

Whatever their provenance, the random juxtaposition of predictive text memes has become fodder for a growing genre of AI humor. Botnik Studios writes goofy songs using souped-up predictive keyboards and a lot of human tweaking. The blog AI Weirdness trains neural nets to do all sorts of ridiculous tasks, such as deciding whether a string of words is more likely to be a name from My Little Pony or a metal band. Darth Vader? 19 percent metal, 81 percent pony. Leia Organa? 96 percent metal, 4 percent pony. (I’m suddenly interpreting Star Wars in quite a new light.)

The combination of the customization and the randomness of the predictive text meme is compelling the way a BuzzFeed quiz or a horoscope is compelling—it gives you a tiny amount of insight into yourself to share, but not so much that you’re baring your soul. It’s also hard to get a truly terrible answer. In both cases, that’s by design.

You know how when you get a new phone and you have to teach it that, no, you aren’t trying to type “duck” and “ducking” all the time? Your keyboard deliberately errs on the conservative side. There are certain words that it just won’t try to complete, even if you get really close. After all, it’s better to accidentally send the word “public” when you meant “pubic” than the other way around.

This goes for sequences of words as well. Just because a sequence is common doesn’t mean it’s a good idea to predict it. “For a while, when you typed ‘I’m going to my Grandma’s,’ GBoard would actually suggest ‘funeral,'” van Esch says. “It’s not wrong, per se. Maybe this is more common than ‘my Grandma’s rave party.’ But at the same time, it’s not something that you want to be reminded about. So it’s better to be a bit careful.”

Users seem to prefer this discretion. Keyboards get roundly criticized when a sexual, morbid, or otherwise disturbing phrase does get predicted. It’s likely that a lot more filtering happens behind the scenes before we even notice it. Janelle Shane, the creator of AI Weirdness, experiences lapses in machine judgment all the time. “Whenever I produce an AI experiment, I’m definitely filtering out offensive content, even when the training data is as innocuous as My Little Pony names. There’s no text-generating algorithm I would trust not to be offensive at some point.”

The true goal of text prediction can’t be as simple as anticipating what a user might want to type. After all, people often type things about sex or death—according to Google Ngrams, “job” is the most common noun after “blow,” and “bucket” is very common after “kick the.” But I experimentally typed these and similar taboo-but-common phrases into my phone’s keyboard, and it never predicted them straightaway. It waited until I’d typed most of the letters of the final word, until I’d definitely committed to the taboo, rather than reminding me of weighty topics when I wasn’t necessarily already thinking about them. With innocuous idioms (like “raining cats and”), the keyboard seemed more proactive about predicting them.

Instead, the goal of text prediction must be to anticipate what the user might want the machine to think they might want to type. For mundane topics, these two goals might seem identical, but their difference shows up as soon as a hint of controversy enters the picture. Predictive text needs to project an aspirational version of a user’s thoughts, a version that avoids subjects like sex and death even though these might be the most important topics to human existence—quite literally the way we enter and leave the world.

We prefer the keyboard to balance raw statistics against our feelings. Sex Death Phone Keyboard is a pretty good name for my future metal band (and a very bad name for my future pony), but I can’t say I’d actually buy a phone that reminds me of my own mortality when I’m composing a grocery list or suggests innuendos when I’m replying to a work email.

The predictive text meme is comforting in a social media world that often leaps from one dismal news cycle to the next. The customizations make us feel seen. The random quirks give our pattern-seeking brains delightful connections. The parts that don’t make sense reassure us of human superiority—the machines can’t be taking over yet if they can’t even write me a decent horoscope! And the topic boundaries prevent the meme from reminding us of our human frailty. The result is a version of ourselves through the verbal equivalent of an Instagram filter, eminently shareable on social media.

More Great WIRED Stories

Tech

[lazilysillyprince]

Autocomplete Presents the Best Version of You

New Post has been published on http://webhostingtop3.com/autocomplete-presents-the-best-version-of-you/

Autocomplete Presents the Best Version of You

Type the phrase “In 2019, I’ll …” and let your smart­phone’s keyboard predict the rest. Depending on what else you’ve typed recently, you might end up with a result like one of these:

In 2019, I’ll let it be a surprise to be honest. In 2019, i’ll be alone. In 2019, I’ll be in the memes of the moment. In 2019, I’ll have to go to get the dog. In 2019 I will rule over the seven kingdoms or my name is not Aegon Targareon [sic].

Many variants on the predictive text meme—which works for both Android and iOS—can be found on social media. Not interested in predicting your 2019? Try writing your villain origin story by following your phone’s suggestions after typing “Foolish heroes! My true plan is …” Test the strength of your personal brand with “You should follow me on Twitter because …” Or launch your political career with “I am running for president with my running mate, @[3rd Twitter Suggestion], because we …”

Gretchen McCulloch is WIRED’s resident linguist. She’s the cocreator of Lingthusiasm, a podcast that’s enthusiastic about linguistics, and her book Because Internet: Understanding the New Rules of Language is coming out in July 2019 from Penguin.

In eight years, we’ve gone from Damn You Autocorrect to treating the strip of three predicted words as a sort of wacky but charming oracle. But when we try to practice divination by algorithm, we’re doing something more than killing a few minutes—we’re exploring the limits of what our devices can and cannot do.

Your phone’s keyboard comes with a basic list of words and sequences of words. That’s what powers the basic language features: autocorrect, where a sequence like “rhe” changes to “the” after you type it, and the suggestion strip just above the letters, which contains both completions (if you type “keyb” it might suggest “keyboard”) and next-word predictions (if you type “predictive” it might suggest “text,” “value,” and “analytics”). It’s this predictions feature that we use to generate amusing and slightly nonsensical strings of text—a function that goes beyond its intended purpose of supplying us with a word or two before we go back to tapping them out letter by letter.

The basic reason we get different results is that, as you use your phone, words or sequences of words that you type get added to your personal word list. “For most users, the on-device dictionary ends up containing local place-names, songs they like, and so on,” says Daan van Esch, a technical program manager of Gboard, Google’s keyboard for Android. Or, in the case of the “Aegon Targareon” example, slightly misspelled Game of Thrones characters.

Another factor that helps us get unique results is a slight bias toward predicting less frequent words. “Suggesting a very common word like ‘and’ might be less helpful because it’s short and easy to type,” van Esch says. “So maybe showing a longer word is actually more useful, even if it’s less frequent.” Of course, a longer word is probably going to be more interesting as meme fodder.

Finally, phones seem to choose different paths from the very beginning. Why are some people getting “I’ll be” while others get “I’ll have” or “I’ll let”? That part is probably not very exciting: The default Android keyboard presumably has slightly different predictions than the default iPhone keyboard, and third-party apps would also have slightly different predictions.

Whatever their provenance, the random juxtaposition of predictive text memes has become fodder for a growing genre of AI humor. Botnik Studios writes goofy songs using souped-up predictive keyboards and a lot of human tweaking. The blog AI Weirdness trains neural nets to do all sorts of ridiculous tasks, such as deciding whether a string of words is more likely to be a name from My Little Pony or a metal band. Darth Vader? 19 percent metal, 81 percent pony. Leia Organa? 96 percent metal, 4 percent pony. (I’m suddenly interpreting Star Wars in quite a new light.)

The combination of the customization and the randomness of the predictive text meme is compelling the way a BuzzFeed quiz or a horoscope is compelling—it gives you a tiny amount of insight into yourself to share, but not so much that you’re baring your soul. It’s also hard to get a truly terrible answer. In both cases, that’s by design.

You know how when you get a new phone and you have to teach it that, no, you aren’t trying to type “duck” and “ducking” all the time? Your keyboard deliberately errs on the conservative side. There are certain words that it just won’t try to complete, even if you get really close. After all, it’s better to accidentally send the word “public” when you meant “pubic” than the other way around.

This goes for sequences of words as well. Just because a sequence is common doesn’t mean it’s a good idea to predict it. “For a while, when you typed ‘I’m going to my Grandma’s,’ GBoard would actually suggest ‘funeral,'” van Esch says. “It’s not wrong, per se. Maybe this is more common than ‘my Grandma’s rave party.’ But at the same time, it’s not something that you want to be reminded about. So it’s better to be a bit careful.”

Users seem to prefer this discretion. Keyboards get roundly criticized when a sexual, morbid, or otherwise disturbing phrase does get predicted. It’s likely that a lot more filtering happens behind the scenes before we even notice it. Janelle Shane, the creator of AI Weirdness, experiences lapses in machine judgment all the time. “Whenever I produce an AI experiment, I’m definitely filtering out offensive content, even when the training data is as innocuous as My Little Pony names. There’s no text-generating algorithm I would trust not to be offensive at some point.”

The true goal of text prediction can’t be as simple as anticipating what a user might want to type. After all, people often type things about sex or death—according to Google Ngrams, “job” is the most common noun after “blow,” and “bucket” is very common after “kick the.” But I experimentally typed these and similar taboo-but-common phrases into my phone’s keyboard, and it never predicted them straightaway. It waited until I’d typed most of the letters of the final word, until I’d definitely committed to the taboo, rather than reminding me of weighty topics when I wasn’t necessarily already thinking about them. With innocuous idioms (like “raining cats and”), the keyboard seemed more proactive about predicting them.

Instead, the goal of text prediction must be to anticipate what the user might want the machine to think they might want to type. For mundane topics, these two goals might seem identical, but their difference shows up as soon as a hint of controversy enters the picture. Predictive text needs to project an aspirational version of a user’s thoughts, a version that avoids subjects like sex and death even though these might be the most important topics to human existence—quite literally the way we enter and leave the world.

We prefer the keyboard to balance raw statistics against our feelings. Sex Death Phone Keyboard is a pretty good name for my future metal band (and a very bad name for my future pony), but I can’t say I’d actually buy a phone that reminds me of my own mortality when I’m composing a grocery list or suggests innuendos when I’m replying to a work email.

The predictive text meme is comforting in a social media world that often leaps from one dismal news cycle to the next. The customizations make us feel seen. The random quirks give our pattern-seeking brains delightful connections. The parts that don’t make sense reassure us of human superiority—the machines can’t be taking over yet if they can’t even write me a decent horoscope! And the topic boundaries prevent the meme from reminding us of our human frailty. The result is a version of ourselves through the verbal equivalent of an Instagram filter, eminently shareable on social media.

More Great WIRED Stories

Tech

[hostingnewsfeed]

Autocomplete Presents the Best Version of You

New Post has been published on http://webhostingtop3.com/autocomplete-presents-the-best-version-of-you/

Autocomplete Presents the Best Version of You

Type the phrase “In 2019, I’ll …” and let your smart­phone’s keyboard predict the rest. Depending on what else you’ve typed recently, you might end up with a result like one of these:

In 2019, I’ll let it be a surprise to be honest. In 2019, i’ll be alone. In 2019, I’ll be in the memes of the moment. In 2019, I’ll have to go to get the dog. In 2019 I will rule over the seven kingdoms or my name is not Aegon Targareon [sic].

Many variants on the predictive text meme—which works for both Android and iOS—can be found on social media. Not interested in predicting your 2019? Try writing your villain origin story by following your phone’s suggestions after typing “Foolish heroes! My true plan is …” Test the strength of your personal brand with “You should follow me on Twitter because …” Or launch your political career with “I am running for president with my running mate, @[3rd Twitter Suggestion], because we …”

Gretchen McCulloch is WIRED’s resident linguist. She’s the cocreator of Lingthusiasm, a podcast that’s enthusiastic about linguistics, and her book Because Internet: Understanding the New Rules of Language is coming out in July 2019 from Penguin.

In eight years, we’ve gone from Damn You Autocorrect to treating the strip of three predicted words as a sort of wacky but charming oracle. But when we try to practice divination by algorithm, we’re doing something more than killing a few minutes—we’re exploring the limits of what our devices can and cannot do.

Your phone’s keyboard comes with a basic list of words and sequences of words. That’s what powers the basic language features: autocorrect, where a sequence like “rhe” changes to “the” after you type it, and the suggestion strip just above the letters, which contains both completions (if you type “keyb” it might suggest “keyboard”) and next-word predictions (if you type “predictive” it might suggest “text,” “value,” and “analytics”). It’s this predictions feature that we use to generate amusing and slightly nonsensical strings of text—a function that goes beyond its intended purpose of supplying us with a word or two before we go back to tapping them out letter by letter.

The basic reason we get different results is that, as you use your phone, words or sequences of words that you type get added to your personal word list. “For most users, the on-device dictionary ends up containing local place-names, songs they like, and so on,” says Daan van Esch, a technical program manager of Gboard, Google’s keyboard for Android. Or, in the case of the “Aegon Targareon” example, slightly misspelled Game of Thrones characters.

Another factor that helps us get unique results is a slight bias toward predicting less frequent words. “Suggesting a very common word like ‘and’ might be less helpful because it’s short and easy to type,” van Esch says. “So maybe showing a longer word is actually more useful, even if it’s less frequent.” Of course, a longer word is probably going to be more interesting as meme fodder.

Finally, phones seem to choose different paths from the very beginning. Why are some people getting “I’ll be” while others get “I’ll have” or “I’ll let”? That part is probably not very exciting: The default Android keyboard presumably has slightly different predictions than the default iPhone keyboard, and third-party apps would also have slightly different predictions.

Whatever their provenance, the random juxtaposition of predictive text memes has become fodder for a growing genre of AI humor. Botnik Studios writes goofy songs using souped-up predictive keyboards and a lot of human tweaking. The blog AI Weirdness trains neural nets to do all sorts of ridiculous tasks, such as deciding whether a string of words is more likely to be a name from My Little Pony or a metal band. Darth Vader? 19 percent metal, 81 percent pony. Leia Organa? 96 percent metal, 4 percent pony. (I’m suddenly interpreting Star Wars in quite a new light.)

The combination of the customization and the randomness of the predictive text meme is compelling the way a BuzzFeed quiz or a horoscope is compelling—it gives you a tiny amount of insight into yourself to share, but not so much that you’re baring your soul. It’s also hard to get a truly terrible answer. In both cases, that’s by design.

You know how when you get a new phone and you have to teach it that, no, you aren’t trying to type “duck” and “ducking” all the time? Your keyboard deliberately errs on the conservative side. There are certain words that it just won’t try to complete, even if you get really close. After all, it’s better to accidentally send the word “public” when you meant “pubic” than the other way around.

This goes for sequences of words as well. Just because a sequence is common doesn’t mean it’s a good idea to predict it. “For a while, when you typed ‘I’m going to my Grandma’s,’ GBoard would actually suggest ‘funeral,'” van Esch says. “It’s not wrong, per se. Maybe this is more common than ‘my Grandma’s rave party.’ But at the same time, it’s not something that you want to be reminded about. So it’s better to be a bit careful.”

Users seem to prefer this discretion. Keyboards get roundly criticized when a sexual, morbid, or otherwise disturbing phrase does get predicted. It’s likely that a lot more filtering happens behind the scenes before we even notice it. Janelle Shane, the creator of AI Weirdness, experiences lapses in machine judgment all the time. “Whenever I produce an AI experiment, I’m definitely filtering out offensive content, even when the training data is as innocuous as My Little Pony names. There’s no text-generating algorithm I would trust not to be offensive at some point.”

The true goal of text prediction can’t be as simple as anticipating what a user might want to type. After all, people often type things about sex or death—according to Google Ngrams, “job” is the most common noun after “blow,” and “bucket” is very common after “kick the.” But I experimentally typed these and similar taboo-but-common phrases into my phone’s keyboard, and it never predicted them straightaway. It waited until I’d typed most of the letters of the final word, until I’d definitely committed to the taboo, rather than reminding me of weighty topics when I wasn’t necessarily already thinking about them. With innocuous idioms (like “raining cats and”), the keyboard seemed more proactive about predicting them.

Instead, the goal of text prediction must be to anticipate what the user might want the machine to think they might want to type. For mundane topics, these two goals might seem identical, but their difference shows up as soon as a hint of controversy enters the picture. Predictive text needs to project an aspirational version of a user’s thoughts, a version that avoids subjects like sex and death even though these might be the most important topics to human existence—quite literally the way we enter and leave the world.

We prefer the keyboard to balance raw statistics against our feelings. Sex Death Phone Keyboard is a pretty good name for my future metal band (and a very bad name for my future pony), but I can’t say I’d actually buy a phone that reminds me of my own mortality when I’m composing a grocery list or suggests innuendos when I’m replying to a work email.

The predictive text meme is comforting in a social media world that often leaps from one dismal news cycle to the next. The customizations make us feel seen. The random quirks give our pattern-seeking brains delightful connections. The parts that don’t make sense reassure us of human superiority—the machines can’t be taking over yet if they can’t even write me a decent horoscope! And the topic boundaries prevent the meme from reminding us of our human frailty. The result is a version of ourselves through the verbal equivalent of an Instagram filter, eminently shareable on social media.

More Great WIRED Stories

Tech

[smartwebhostingblog]

Autocomplete Presents the Best Version of You

New Post has been published on http://webhostingtop3.com/autocomplete-presents-the-best-version-of-you/

Autocomplete Presents the Best Version of You

Type the phrase “In 2019, I’ll …” and let your smart­phone’s keyboard predict the rest. Depending on what else you’ve typed recently, you might end up with a result like one of these:

In 2019, I’ll let it be a surprise to be honest. In 2019, i’ll be alone. In 2019, I’ll be in the memes of the moment. In 2019, I’ll have to go to get the dog. In 2019 I will rule over the seven kingdoms or my name is not Aegon Targareon [sic].

Many variants on the predictive text meme—which works for both Android and iOS—can be found on social media. Not interested in predicting your 2019? Try writing your villain origin story by following your phone’s suggestions after typing “Foolish heroes! My true plan is …” Test the strength of your personal brand with “You should follow me on Twitter because …” Or launch your political career with “I am running for president with my running mate, @[3rd Twitter Suggestion], because we …”

Gretchen McCulloch is WIRED’s resident linguist. She’s the cocreator of Lingthusiasm, a podcast that’s enthusiastic about linguistics, and her book Because Internet: Understanding the New Rules of Language is coming out in July 2019 from Penguin.

In eight years, we’ve gone from Damn You Autocorrect to treating the strip of three predicted words as a sort of wacky but charming oracle. But when we try to practice divination by algorithm, we’re doing something more than killing a few minutes—we’re exploring the limits of what our devices can and cannot do.

Your phone’s keyboard comes with a basic list of words and sequences of words. That’s what powers the basic language features: autocorrect, where a sequence like “rhe” changes to “the” after you type it, and the suggestion strip just above the letters, which contains both completions (if you type “keyb” it might suggest “keyboard”) and next-word predictions (if you type “predictive” it might suggest “text,” “value,” and “analytics”). It’s this predictions feature that we use to generate amusing and slightly nonsensical strings of text—a function that goes beyond its intended purpose of supplying us with a word or two before we go back to tapping them out letter by letter.

The basic reason we get different results is that, as you use your phone, words or sequences of words that you type get added to your personal word list. “For most users, the on-device dictionary ends up containing local place-names, songs they like, and so on,” says Daan van Esch, a technical program manager of Gboard, Google’s keyboard for Android. Or, in the case of the “Aegon Targareon” example, slightly misspelled Game of Thrones characters.

Another factor that helps us get unique results is a slight bias toward predicting less frequent words. “Suggesting a very common word like ‘and’ might be less helpful because it’s short and easy to type,” van Esch says. “So maybe showing a longer word is actually more useful, even if it’s less frequent.” Of course, a longer word is probably going to be more interesting as meme fodder.

Finally, phones seem to choose different paths from the very beginning. Why are some people getting “I’ll be” while others get “I’ll have” or “I’ll let”? That part is probably not very exciting: The default Android keyboard presumably has slightly different predictions than the default iPhone keyboard, and third-party apps would also have slightly different predictions.

Whatever their provenance, the random juxtaposition of predictive text memes has become fodder for a growing genre of AI humor. Botnik Studios writes goofy songs using souped-up predictive keyboards and a lot of human tweaking. The blog AI Weirdness trains neural nets to do all sorts of ridiculous tasks, such as deciding whether a string of words is more likely to be a name from My Little Pony or a metal band. Darth Vader? 19 percent metal, 81 percent pony. Leia Organa? 96 percent metal, 4 percent pony. (I’m suddenly interpreting Star Wars in quite a new light.)

The combination of the customization and the randomness of the predictive text meme is compelling the way a BuzzFeed quiz or a horoscope is compelling—it gives you a tiny amount of insight into yourself to share, but not so much that you’re baring your soul. It’s also hard to get a truly terrible answer. In both cases, that’s by design.

You know how when you get a new phone and you have to teach it that, no, you aren’t trying to type “duck” and “ducking” all the time? Your keyboard deliberately errs on the conservative side. There are certain words that it just won’t try to complete, even if you get really close. After all, it’s better to accidentally send the word “public” when you meant “pubic” than the other way around.

This goes for sequences of words as well. Just because a sequence is common doesn’t mean it’s a good idea to predict it. “For a while, when you typed ‘I’m going to my Grandma’s,’ GBoard would actually suggest ‘funeral,'” van Esch says. “It’s not wrong, per se. Maybe this is more common than ‘my Grandma’s rave party.’ But at the same time, it’s not something that you want to be reminded about. So it’s better to be a bit careful.”

Users seem to prefer this discretion. Keyboards get roundly criticized when a sexual, morbid, or otherwise disturbing phrase does get predicted. It’s likely that a lot more filtering happens behind the scenes before we even notice it. Janelle Shane, the creator of AI Weirdness, experiences lapses in machine judgment all the time. “Whenever I produce an AI experiment, I’m definitely filtering out offensive content, even when the training data is as innocuous as My Little Pony names. There’s no text-generating algorithm I would trust not to be offensive at some point.”

The true goal of text prediction can’t be as simple as anticipating what a user might want to type. After all, people often type things about sex or death—according to Google Ngrams, “job” is the most common noun after “blow,” and “bucket” is very common after “kick the.” But I experimentally typed these and similar taboo-but-common phrases into my phone’s keyboard, and it never predicted them straightaway. It waited until I’d typed most of the letters of the final word, until I’d definitely committed to the taboo, rather than reminding me of weighty topics when I wasn’t necessarily already thinking about them. With innocuous idioms (like “raining cats and”), the keyboard seemed more proactive about predicting them.

Instead, the goal of text prediction must be to anticipate what the user might want the machine to think they might want to type. For mundane topics, these two goals might seem identical, but their difference shows up as soon as a hint of controversy enters the picture. Predictive text needs to project an aspirational version of a user’s thoughts, a version that avoids subjects like sex and death even though these might be the most important topics to human existence—quite literally the way we enter and leave the world.

We prefer the keyboard to balance raw statistics against our feelings. Sex Death Phone Keyboard is a pretty good name for my future metal band (and a very bad name for my future pony), but I can’t say I’d actually buy a phone that reminds me of my own mortality when I’m composing a grocery list or suggests innuendos when I’m replying to a work email.

The predictive text meme is comforting in a social media world that often leaps from one dismal news cycle to the next. The customizations make us feel seen. The random quirks give our pattern-seeking brains delightful connections. The parts that don’t make sense reassure us of human superiority—the machines can’t be taking over yet if they can’t even write me a decent horoscope! And the topic boundaries prevent the meme from reminding us of our human frailty. The result is a version of ourselves through the verbal equivalent of an Instagram filter, eminently shareable on social media.

More Great WIRED Stories

Tech

[smartwebhostingblog]

Autocomplete Presents the Best Version of You

New Post has been published on http://webhostingtop3.com/autocomplete-presents-the-best-version-of-you/

Autocomplete Presents the Best Version of You

Type the phrase “In 2019, I’ll …” and let your smart­phone’s keyboard predict the rest. Depending on what else you’ve typed recently, you might end up with a result like one of these:

In 2019, I’ll let it be a surprise to be honest. In 2019, i’ll be alone. In 2019, I’ll be in the memes of the moment. In 2019, I’ll have to go to get the dog. In 2019 I will rule over the seven kingdoms or my name is not Aegon Targareon [sic].

Many variants on the predictive text meme—which works for both Android and iOS—can be found on social media. Not interested in predicting your 2019? Try writing your villain origin story by following your phone’s suggestions after typing “Foolish heroes! My true plan is …” Test the strength of your personal brand with “You should follow me on Twitter because …” Or launch your political career with “I am running for president with my running mate, @[3rd Twitter Suggestion], because we …”

Gretchen McCulloch is WIRED’s resident linguist. She’s the cocreator of Lingthusiasm, a podcast that’s enthusiastic about linguistics, and her book Because Internet: Understanding the New Rules of Language is coming out in July 2019 from Penguin.

In eight years, we’ve gone from Damn You Autocorrect to treating the strip of three predicted words as a sort of wacky but charming oracle. But when we try to practice divination by algorithm, we’re doing something more than killing a few minutes—we’re exploring the limits of what our devices can and cannot do.

Your phone’s keyboard comes with a basic list of words and sequences of words. That’s what powers the basic language features: autocorrect, where a sequence like “rhe” changes to “the” after you type it, and the suggestion strip just above the letters, which contains both completions (if you type “keyb” it might suggest “keyboard”) and next-word predictions (if you type “predictive” it might suggest “text,” “value,” and “analytics”). It’s this predictions feature that we use to generate amusing and slightly nonsensical strings of text—a function that goes beyond its intended purpose of supplying us with a word or two before we go back to tapping them out letter by letter.

The basic reason we get different results is that, as you use your phone, words or sequences of words that you type get added to your personal word list. “For most users, the on-device dictionary ends up containing local place-names, songs they like, and so on,” says Daan van Esch, a technical program manager of Gboard, Google’s keyboard for Android. Or, in the case of the “Aegon Targareon” example, slightly misspelled Game of Thrones characters.

Another factor that helps us get unique results is a slight bias toward predicting less frequent words. “Suggesting a very common word like ‘and’ might be less helpful because it’s short and easy to type,” van Esch says. “So maybe showing a longer word is actually more useful, even if it’s less frequent.” Of course, a longer word is probably going to be more interesting as meme fodder.

Finally, phones seem to choose different paths from the very beginning. Why are some people getting “I’ll be” while others get “I’ll have” or “I’ll let”? That part is probably not very exciting: The default Android keyboard presumably has slightly different predictions than the default iPhone keyboard, and third-party apps would also have slightly different predictions.

Whatever their provenance, the random juxtaposition of predictive text memes has become fodder for a growing genre of AI humor. Botnik Studios writes goofy songs using souped-up predictive keyboards and a lot of human tweaking. The blog AI Weirdness trains neural nets to do all sorts of ridiculous tasks, such as deciding whether a string of words is more likely to be a name from My Little Pony or a metal band. Darth Vader? 19 percent metal, 81 percent pony. Leia Organa? 96 percent metal, 4 percent pony. (I’m suddenly interpreting Star Wars in quite a new light.)

The combination of the customization and the randomness of the predictive text meme is compelling the way a BuzzFeed quiz or a horoscope is compelling—it gives you a tiny amount of insight into yourself to share, but not so much that you’re baring your soul. It’s also hard to get a truly terrible answer. In both cases, that’s by design.

You know how when you get a new phone and you have to teach it that, no, you aren’t trying to type “duck” and “ducking” all the time? Your keyboard deliberately errs on the conservative side. There are certain words that it just won’t try to complete, even if you get really close. After all, it’s better to accidentally send the word “public” when you meant “pubic” than the other way around.

This goes for sequences of words as well. Just because a sequence is common doesn’t mean it’s a good idea to predict it. “For a while, when you typed ‘I’m going to my Grandma’s,’ GBoard would actually suggest ‘funeral,'” van Esch says. “It’s not wrong, per se. Maybe this is more common than ‘my Grandma’s rave party.’ But at the same time, it’s not something that you want to be reminded about. So it’s better to be a bit careful.”

Users seem to prefer this discretion. Keyboards get roundly criticized when a sexual, morbid, or otherwise disturbing phrase does get predicted. It’s likely that a lot more filtering happens behind the scenes before we even notice it. Janelle Shane, the creator of AI Weirdness, experiences lapses in machine judgment all the time. “Whenever I produce an AI experiment, I’m definitely filtering out offensive content, even when the training data is as innocuous as My Little Pony names. There’s no text-generating algorithm I would trust not to be offensive at some point.”

The true goal of text prediction can’t be as simple as anticipating what a user might want to type. After all, people often type things about sex or death—according to Google Ngrams, “job” is the most common noun after “blow,” and “bucket” is very common after “kick the.” But I experimentally typed these and similar taboo-but-common phrases into my phone’s keyboard, and it never predicted them straightaway. It waited until I’d typed most of the letters of the final word, until I’d definitely committed to the taboo, rather than reminding me of weighty topics when I wasn’t necessarily already thinking about them. With innocuous idioms (like “raining cats and”), the keyboard seemed more proactive about predicting them.

Instead, the goal of text prediction must be to anticipate what the user might want the machine to think they might want to type. For mundane topics, these two goals might seem identical, but their difference shows up as soon as a hint of controversy enters the picture. Predictive text needs to project an aspirational version of a user’s thoughts, a version that avoids subjects like sex and death even though these might be the most important topics to human existence—quite literally the way we enter and leave the world.

We prefer the keyboard to balance raw statistics against our feelings. Sex Death Phone Keyboard is a pretty good name for my future metal band (and a very bad name for my future pony), but I can’t say I’d actually buy a phone that reminds me of my own mortality when I’m composing a grocery list or suggests innuendos when I’m replying to a work email.

The predictive text meme is comforting in a social media world that often leaps from one dismal news cycle to the next. The customizations make us feel seen. The random quirks give our pattern-seeking brains delightful connections. The parts that don’t make sense reassure us of human superiority—the machines can’t be taking over yet if they can’t even write me a decent horoscope! And the topic boundaries prevent the meme from reminding us of our human frailty. The result is a version of ourselves through the verbal equivalent of an Instagram filter, eminently shareable on social media.

More Great WIRED Stories

Tech

[awesomecolorfulcollectorwombat]

Autocomplete Presents the Best Version of You

New Post has been published on http://webhostingtop3.com/autocomplete-presents-the-best-version-of-you/

Autocomplete Presents the Best Version of You

Type the phrase “In 2019, I’ll …” and let your smart­phone’s keyboard predict the rest. Depending on what else you’ve typed recently, you might end up with a result like one of these:

In 2019, I’ll let it be a surprise to be honest. In 2019, i’ll be alone. In 2019, I’ll be in the memes of the moment. In 2019, I’ll have to go to get the dog. In 2019 I will rule over the seven kingdoms or my name is not Aegon Targareon [sic].

Many variants on the predictive text meme—which works for both Android and iOS—can be found on social media. Not interested in predicting your 2019? Try writing your villain origin story by following your phone’s suggestions after typing “Foolish heroes! My true plan is …” Test the strength of your personal brand with “You should follow me on Twitter because …” Or launch your political career with “I am running for president with my running mate, @[3rd Twitter Suggestion], because we …”

Gretchen McCulloch is WIRED’s resident linguist. She’s the cocreator of Lingthusiasm, a podcast that’s enthusiastic about linguistics, and her book Because Internet: Understanding the New Rules of Language is coming out in July 2019 from Penguin.

In eight years, we’ve gone from Damn You Autocorrect to treating the strip of three predicted words as a sort of wacky but charming oracle. But when we try to practice divination by algorithm, we’re doing something more than killing a few minutes—we’re exploring the limits of what our devices can and cannot do.

Your phone’s keyboard comes with a basic list of words and sequences of words. That’s what powers the basic language features: autocorrect, where a sequence like “rhe” changes to “the” after you type it, and the suggestion strip just above the letters, which contains both completions (if you type “keyb” it might suggest “keyboard”) and next-word predictions (if you type “predictive” it might suggest “text,” “value,” and “analytics”). It’s this predictions feature that we use to generate amusing and slightly nonsensical strings of text—a function that goes beyond its intended purpose of supplying us with a word or two before we go back to tapping them out letter by letter.

The basic reason we get different results is that, as you use your phone, words or sequences of words that you type get added to your personal word list. “For most users, the on-device dictionary ends up containing local place-names, songs they like, and so on,” says Daan van Esch, a technical program manager of Gboard, Google’s keyboard for Android. Or, in the case of the “Aegon Targareon” example, slightly misspelled Game of Thrones characters.

Another factor that helps us get unique results is a slight bias toward predicting less frequent words. “Suggesting a very common word like ‘and’ might be less helpful because it’s short and easy to type,” van Esch says. “So maybe showing a longer word is actually more useful, even if it’s less frequent.” Of course, a longer word is probably going to be more interesting as meme fodder.

Finally, phones seem to choose different paths from the very beginning. Why are some people getting “I’ll be” while others get “I’ll have” or “I’ll let”? That part is probably not very exciting: The default Android keyboard presumably has slightly different predictions than the default iPhone keyboard, and third-party apps would also have slightly different predictions.

Whatever their provenance, the random juxtaposition of predictive text memes has become fodder for a growing genre of AI humor. Botnik Studios writes goofy songs using souped-up predictive keyboards and a lot of human tweaking. The blog AI Weirdness trains neural nets to do all sorts of ridiculous tasks, such as deciding whether a string of words is more likely to be a name from My Little Pony or a metal band. Darth Vader? 19 percent metal, 81 percent pony. Leia Organa? 96 percent metal, 4 percent pony. (I’m suddenly interpreting Star Wars in quite a new light.)

The combination of the customization and the randomness of the predictive text meme is compelling the way a BuzzFeed quiz or a horoscope is compelling—it gives you a tiny amount of insight into yourself to share, but not so much that you’re baring your soul. It’s also hard to get a truly terrible answer. In both cases, that’s by design.

You know how when you get a new phone and you have to teach it that, no, you aren’t trying to type “duck” and “ducking” all the time? Your keyboard deliberately errs on the conservative side. There are certain words that it just won’t try to complete, even if you get really close. After all, it’s better to accidentally send the word “public” when you meant “pubic” than the other way around.

This goes for sequences of words as well. Just because a sequence is common doesn’t mean it’s a good idea to predict it. “For a while, when you typed ‘I’m going to my Grandma’s,’ GBoard would actually suggest ‘funeral,'” van Esch says. “It’s not wrong, per se. Maybe this is more common than ‘my Grandma’s rave party.’ But at the same time, it’s not something that you want to be reminded about. So it’s better to be a bit careful.”

Users seem to prefer this discretion. Keyboards get roundly criticized when a sexual, morbid, or otherwise disturbing phrase does get predicted. It’s likely that a lot more filtering happens behind the scenes before we even notice it. Janelle Shane, the creator of AI Weirdness, experiences lapses in machine judgment all the time. “Whenever I produce an AI experiment, I’m definitely filtering out offensive content, even when the training data is as innocuous as My Little Pony names. There’s no text-generating algorithm I would trust not to be offensive at some point.”

The true goal of text prediction can’t be as simple as anticipating what a user might want to type. After all, people often type things about sex or death—according to Google Ngrams, “job” is the most common noun after “blow,” and “bucket” is very common after “kick the.” But I experimentally typed these and similar taboo-but-common phrases into my phone’s keyboard, and it never predicted them straightaway. It waited until I’d typed most of the letters of the final word, until I’d definitely committed to the taboo, rather than reminding me of weighty topics when I wasn’t necessarily already thinking about them. With innocuous idioms (like “raining cats and”), the keyboard seemed more proactive about predicting them.

Instead, the goal of text prediction must be to anticipate what the user might want the machine to think they might want to type. For mundane topics, these two goals might seem identical, but their difference shows up as soon as a hint of controversy enters the picture. Predictive text needs to project an aspirational version of a user’s thoughts, a version that avoids subjects like sex and death even though these might be the most important topics to human existence—quite literally the way we enter and leave the world.

We prefer the keyboard to balance raw statistics against our feelings. Sex Death Phone Keyboard is a pretty good name for my future metal band (and a very bad name for my future pony), but I can’t say I’d actually buy a phone that reminds me of my own mortality when I’m composing a grocery list or suggests innuendos when I’m replying to a work email.

The predictive text meme is comforting in a social media world that often leaps from one dismal news cycle to the next. The customizations make us feel seen. The random quirks give our pattern-seeking brains delightful connections. The parts that don’t make sense reassure us of human superiority—the machines can’t be taking over yet if they can’t even write me a decent horoscope! And the topic boundaries prevent the meme from reminding us of our human frailty. The result is a version of ourselves through the verbal equivalent of an Instagram filter, eminently shareable on social media.

More Great WIRED Stories

Tech

[awesomecolorfulcollectorwombat]

Autocomplete Presents the Best Version of You

New Post has been published on http://webhostingtop3.com/autocomplete-presents-the-best-version-of-you/

Autocomplete Presents the Best Version of You

Type the phrase “In 2019, I’ll …” and let your smart­phone’s keyboard predict the rest. Depending on what else you’ve typed recently, you might end up with a result like one of these:

In 2019, I’ll let it be a surprise to be honest. In 2019, i’ll be alone. In 2019, I’ll be in the memes of the moment. In 2019, I’ll have to go to get the dog. In 2019 I will rule over the seven kingdoms or my name is not Aegon Targareon [sic].

Many variants on the predictive text meme—which works for both Android and iOS—can be found on social media. Not interested in predicting your 2019? Try writing your villain origin story by following your phone’s suggestions after typing “Foolish heroes! My true plan is …” Test the strength of your personal brand with “You should follow me on Twitter because …” Or launch your political career with “I am running for president with my running mate, @[3rd Twitter Suggestion], because we …”

Gretchen McCulloch is WIRED’s resident linguist. She’s the cocreator of Lingthusiasm, a podcast that’s enthusiastic about linguistics, and her book Because Internet: Understanding the New Rules of Language is coming out in July 2019 from Penguin.

In eight years, we’ve gone from Damn You Autocorrect to treating the strip of three predicted words as a sort of wacky but charming oracle. But when we try to practice divination by algorithm, we’re doing something more than killing a few minutes—we’re exploring the limits of what our devices can and cannot do.

Your phone’s keyboard comes with a basic list of words and sequences of words. That’s what powers the basic language features: autocorrect, where a sequence like “rhe” changes to “the” after you type it, and the suggestion strip just above the letters, which contains both completions (if you type “keyb” it might suggest “keyboard”) and next-word predictions (if you type “predictive” it might suggest “text,” “value,” and “analytics”). It’s this predictions feature that we use to generate amusing and slightly nonsensical strings of text—a function that goes beyond its intended purpose of supplying us with a word or two before we go back to tapping them out letter by letter.

The basic reason we get different results is that, as you use your phone, words or sequences of words that you type get added to your personal word list. “For most users, the on-device dictionary ends up containing local place-names, songs they like, and so on,” says Daan van Esch, a technical program manager of Gboard, Google’s keyboard for Android. Or, in the case of the “Aegon Targareon” example, slightly misspelled Game of Thrones characters.

Another factor that helps us get unique results is a slight bias toward predicting less frequent words. “Suggesting a very common word like ‘and’ might be less helpful because it’s short and easy to type,” van Esch says. “So maybe showing a longer word is actually more useful, even if it’s less frequent.” Of course, a longer word is probably going to be more interesting as meme fodder.

Finally, phones seem to choose different paths from the very beginning. Why are some people getting “I’ll be” while others get “I’ll have” or “I’ll let”? That part is probably not very exciting: The default Android keyboard presumably has slightly different predictions than the default iPhone keyboard, and third-party apps would also have slightly different predictions.

Whatever their provenance, the random juxtaposition of predictive text memes has become fodder for a growing genre of AI humor. Botnik Studios writes goofy songs using souped-up predictive keyboards and a lot of human tweaking. The blog AI Weirdness trains neural nets to do all sorts of ridiculous tasks, such as deciding whether a string of words is more likely to be a name from My Little Pony or a metal band. Darth Vader? 19 percent metal, 81 percent pony. Leia Organa? 96 percent metal, 4 percent pony. (I’m suddenly interpreting Star Wars in quite a new light.)

The combination of the customization and the randomness of the predictive text meme is compelling the way a BuzzFeed quiz or a horoscope is compelling—it gives you a tiny amount of insight into yourself to share, but not so much that you’re baring your soul. It’s also hard to get a truly terrible answer. In both cases, that’s by design.

You know how when you get a new phone and you have to teach it that, no, you aren’t trying to type “duck” and “ducking” all the time? Your keyboard deliberately errs on the conservative side. There are certain words that it just won’t try to complete, even if you get really close. After all, it’s better to accidentally send the word “public” when you meant “pubic” than the other way around.

This goes for sequences of words as well. Just because a sequence is common doesn’t mean it’s a good idea to predict it. “For a while, when you typed ‘I’m going to my Grandma’s,’ GBoard would actually suggest ‘funeral,'” van Esch says. “It’s not wrong, per se. Maybe this is more common than ‘my Grandma’s rave party.’ But at the same time, it’s not something that you want to be reminded about. So it’s better to be a bit careful.”

Users seem to prefer this discretion. Keyboards get roundly criticized when a sexual, morbid, or otherwise disturbing phrase does get predicted. It’s likely that a lot more filtering happens behind the scenes before we even notice it. Janelle Shane, the creator of AI Weirdness, experiences lapses in machine judgment all the time. “Whenever I produce an AI experiment, I’m definitely filtering out offensive content, even when the training data is as innocuous as My Little Pony names. There’s no text-generating algorithm I would trust not to be offensive at some point.”

The true goal of text prediction can’t be as simple as anticipating what a user might want to type. After all, people often type things about sex or death—according to Google Ngrams, “job” is the most common noun after “blow,” and “bucket” is very common after “kick the.” But I experimentally typed these and similar taboo-but-common phrases into my phone’s keyboard, and it never predicted them straightaway. It waited until I’d typed most of the letters of the final word, until I’d definitely committed to the taboo, rather than reminding me of weighty topics when I wasn’t necessarily already thinking about them. With innocuous idioms (like “raining cats and”), the keyboard seemed more proactive about predicting them.

Instead, the goal of text prediction must be to anticipate what the user might want the machine to think they might want to type. For mundane topics, these two goals might seem identical, but their difference shows up as soon as a hint of controversy enters the picture. Predictive text needs to project an aspirational version of a user’s thoughts, a version that avoids subjects like sex and death even though these might be the most important topics to human existence—quite literally the way we enter and leave the world.

We prefer the keyboard to balance raw statistics against our feelings. Sex Death Phone Keyboard is a pretty good name for my future metal band (and a very bad name for my future pony), but I can’t say I’d actually buy a phone that reminds me of my own mortality when I’m composing a grocery list or suggests innuendos when I’m replying to a work email.

The predictive text meme is comforting in a social media world that often leaps from one dismal news cycle to the next. The customizations make us feel seen. The random quirks give our pattern-seeking brains delightful connections. The parts that don’t make sense reassure us of human superiority—the machines can’t be taking over yet if they can’t even write me a decent horoscope! And the topic boundaries prevent the meme from reminding us of our human frailty. The result is a version of ourselves through the verbal equivalent of an Instagram filter, eminently shareable on social media.

More Great WIRED Stories

Tech

[lazilysillyprince]

Autocomplete Presents the Best Version of You

New Post has been published on http://webhostingtop3.com/autocomplete-presents-the-best-version-of-you/

Autocomplete Presents the Best Version of You

Type the phrase “In 2019, I’ll …” and let your smart­phone’s keyboard predict the rest. Depending on what else you’ve typed recently, you might end up with a result like one of these:

In 2019, I’ll let it be a surprise to be honest. In 2019, i’ll be alone. In 2019, I’ll be in the memes of the moment. In 2019, I’ll have to go to get the dog. In 2019 I will rule over the seven kingdoms or my name is not Aegon Targareon [sic].

Many variants on the predictive text meme—which works for both Android and iOS—can be found on social media. Not interested in predicting your 2019? Try writing your villain origin story by following your phone’s suggestions after typing “Foolish heroes! My true plan is …” Test the strength of your personal brand with “You should follow me on Twitter because …” Or launch your political career with “I am running for president with my running mate, @[3rd Twitter Suggestion], because we …”

Gretchen McCulloch is WIRED’s resident linguist. She’s the cocreator of Lingthusiasm, a podcast that’s enthusiastic about linguistics, and her book Because Internet: Understanding the New Rules of Language is coming out in July 2019 from Penguin.

In eight years, we’ve gone from Damn You Autocorrect to treating the strip of three predicted words as a sort of wacky but charming oracle. But when we try to practice divination by algorithm, we’re doing something more than killing a few minutes—we’re exploring the limits of what our devices can and cannot do.

Your phone’s keyboard comes with a basic list of words and sequences of words. That’s what powers the basic language features: autocorrect, where a sequence like “rhe” changes to “the” after you type it, and the suggestion strip just above the letters, which contains both completions (if you type “keyb” it might suggest “keyboard”) and next-word predictions (if you type “predictive” it might suggest “text,” “value,” and “analytics”). It’s this predictions feature that we use to generate amusing and slightly nonsensical strings of text—a function that goes beyond its intended purpose of supplying us with a word or two before we go back to tapping them out letter by letter.

The basic reason we get different results is that, as you use your phone, words or sequences of words that you type get added to your personal word list. “For most users, the on-device dictionary ends up containing local place-names, songs they like, and so on,” says Daan van Esch, a technical program manager of Gboard, Google’s keyboard for Android. Or, in the case of the “Aegon Targareon” example, slightly misspelled Game of Thrones characters.

Another factor that helps us get unique results is a slight bias toward predicting less frequent words. “Suggesting a very common word like ‘and’ might be less helpful because it’s short and easy to type,” van Esch says. “So maybe showing a longer word is actually more useful, even if it’s less frequent.” Of course, a longer word is probably going to be more interesting as meme fodder.

Finally, phones seem to choose different paths from the very beginning. Why are some people getting “I’ll be” while others get “I’ll have” or “I’ll let”? That part is probably not very exciting: The default Android keyboard presumably has slightly different predictions than the default iPhone keyboard, and third-party apps would also have slightly different predictions.

Whatever their provenance, the random juxtaposition of predictive text memes has become fodder for a growing genre of AI humor. Botnik Studios writes goofy songs using souped-up predictive keyboards and a lot of human tweaking. The blog AI Weirdness trains neural nets to do all sorts of ridiculous tasks, such as deciding whether a string of words is more likely to be a name from My Little Pony or a metal band. Darth Vader? 19 percent metal, 81 percent pony. Leia Organa? 96 percent metal, 4 percent pony. (I’m suddenly interpreting Star Wars in quite a new light.)

The combination of the customization and the randomness of the predictive text meme is compelling the way a BuzzFeed quiz or a horoscope is compelling—it gives you a tiny amount of insight into yourself to share, but not so much that you’re baring your soul. It’s also hard to get a truly terrible answer. In both cases, that’s by design.

You know how when you get a new phone and you have to teach it that, no, you aren’t trying to type “duck” and “ducking” all the time? Your keyboard deliberately errs on the conservative side. There are certain words that it just won’t try to complete, even if you get really close. After all, it’s better to accidentally send the word “public” when you meant “pubic” than the other way around.

This goes for sequences of words as well. Just because a sequence is common doesn’t mean it’s a good idea to predict it. “For a while, when you typed ‘I’m going to my Grandma’s,’ GBoard would actually suggest ‘funeral,'” van Esch says. “It’s not wrong, per se. Maybe this is more common than ‘my Grandma’s rave party.’ But at the same time, it’s not something that you want to be reminded about. So it’s better to be a bit careful.”

Users seem to prefer this discretion. Keyboards get roundly criticized when a sexual, morbid, or otherwise disturbing phrase does get predicted. It’s likely that a lot more filtering happens behind the scenes before we even notice it. Janelle Shane, the creator of AI Weirdness, experiences lapses in machine judgment all the time. “Whenever I produce an AI experiment, I’m definitely filtering out offensive content, even when the training data is as innocuous as My Little Pony names. There’s no text-generating algorithm I would trust not to be offensive at some point.”

The true goal of text prediction can’t be as simple as anticipating what a user might want to type. After all, people often type things about sex or death—according to Google Ngrams, “job” is the most common noun after “blow,” and “bucket” is very common after “kick the.” But I experimentally typed these and similar taboo-but-common phrases into my phone’s keyboard, and it never predicted them straightaway. It waited until I’d typed most of the letters of the final word, until I’d definitely committed to the taboo, rather than reminding me of weighty topics when I wasn’t necessarily already thinking about them. With innocuous idioms (like “raining cats and”), the keyboard seemed more proactive about predicting them.

Instead, the goal of text prediction must be to anticipate what the user might want the machine to think they might want to type. For mundane topics, these two goals might seem identical, but their difference shows up as soon as a hint of controversy enters the picture. Predictive text needs to project an aspirational version of a user’s thoughts, a version that avoids subjects like sex and death even though these might be the most important topics to human existence—quite literally the way we enter and leave the world.

We prefer the keyboard to balance raw statistics against our feelings. Sex Death Phone Keyboard is a pretty good name for my future metal band (and a very bad name for my future pony), but I can’t say I’d actually buy a phone that reminds me of my own mortality when I’m composing a grocery list or suggests innuendos when I’m replying to a work email.

The predictive text meme is comforting in a social media world that often leaps from one dismal news cycle to the next. The customizations make us feel seen. The random quirks give our pattern-seeking brains delightful connections. The parts that don’t make sense reassure us of human superiority—the machines can’t be taking over yet if they can’t even write me a decent horoscope! And the topic boundaries prevent the meme from reminding us of our human frailty. The result is a version of ourselves through the verbal equivalent of an Instagram filter, eminently shareable on social media.

More Great WIRED Stories

Tech

[lazilysillyprince]

Autocomplete Presents the Best Version of You

New Post has been published on http://webhostingtop3.com/autocomplete-presents-the-best-version-of-you/

Autocomplete Presents the Best Version of You

Type the phrase “In 2019, I’ll …” and let your smart­phone’s keyboard predict the rest. Depending on what else you’ve typed recently, you might end up with a result like one of these:

In 2019, I’ll let it be a surprise to be honest. In 2019, i’ll be alone. In 2019, I’ll be in the memes of the moment. In 2019, I’ll have to go to get the dog. In 2019 I will rule over the seven kingdoms or my name is not Aegon Targareon [sic].

Many variants on the predictive text meme—which works for both Android and iOS—can be found on social media. Not interested in predicting your 2019? Try writing your villain origin story by following your phone’s suggestions after typing “Foolish heroes! My true plan is …” Test the strength of your personal brand with “You should follow me on Twitter because …” Or launch your political career with “I am running for president with my running mate, @[3rd Twitter Suggestion], because we …”

Gretchen McCulloch is WIRED’s resident linguist. She’s the cocreator of Lingthusiasm, a podcast that’s enthusiastic about linguistics, and her book Because Internet: Understanding the New Rules of Language is coming out in July 2019 from Penguin.

In eight years, we’ve gone from Damn You Autocorrect to treating the strip of three predicted words as a sort of wacky but charming oracle. But when we try to practice divination by algorithm, we’re doing something more than killing a few minutes—we’re exploring the limits of what our devices can and cannot do.

Your phone’s keyboard comes with a basic list of words and sequences of words. That’s what powers the basic language features: autocorrect, where a sequence like “rhe” changes to “the” after you type it, and the suggestion strip just above the letters, which contains both completions (if you type “keyb” it might suggest “keyboard”) and next-word predictions (if you type “predictive” it might suggest “text,” “value,” and “analytics”). It’s this predictions feature that we use to generate amusing and slightly nonsensical strings of text—a function that goes beyond its intended purpose of supplying us with a word or two before we go back to tapping them out letter by letter.

The basic reason we get different results is that, as you use your phone, words or sequences of words that you type get added to your personal word list. “For most users, the on-device dictionary ends up containing local place-names, songs they like, and so on,” says Daan van Esch, a technical program manager of Gboard, Google’s keyboard for Android. Or, in the case of the “Aegon Targareon” example, slightly misspelled Game of Thrones characters.

Another factor that helps us get unique results is a slight bias toward predicting less frequent words. “Suggesting a very common word like ‘and’ might be less helpful because it’s short and easy to type,” van Esch says. “So maybe showing a longer word is actually more useful, even if it’s less frequent.” Of course, a longer word is probably going to be more interesting as meme fodder.

Finally, phones seem to choose different paths from the very beginning. Why are some people getting “I’ll be” while others get “I’ll have” or “I’ll let”? That part is probably not very exciting: The default Android keyboard presumably has slightly different predictions than the default iPhone keyboard, and third-party apps would also have slightly different predictions.

Whatever their provenance, the random juxtaposition of predictive text memes has become fodder for a growing genre of AI humor. Botnik Studios writes goofy songs using souped-up predictive keyboards and a lot of human tweaking. The blog AI Weirdness trains neural nets to do all sorts of ridiculous tasks, such as deciding whether a string of words is more likely to be a name from My Little Pony or a metal band. Darth Vader? 19 percent metal, 81 percent pony. Leia Organa? 96 percent metal, 4 percent pony. (I’m suddenly interpreting Star Wars in quite a new light.)

The combination of the customization and the randomness of the predictive text meme is compelling the way a BuzzFeed quiz or a horoscope is compelling—it gives you a tiny amount of insight into yourself to share, but not so much that you’re baring your soul. It’s also hard to get a truly terrible answer. In both cases, that’s by design.

You know how when you get a new phone and you have to teach it that, no, you aren’t trying to type “duck” and “ducking” all the time? Your keyboard deliberately errs on the conservative side. There are certain words that it just won’t try to complete, even if you get really close. After all, it’s better to accidentally send the word “public” when you meant “pubic” than the other way around.

This goes for sequences of words as well. Just because a sequence is common doesn’t mean it’s a good idea to predict it. “For a while, when you typed ‘I’m going to my Grandma’s,’ GBoard would actually suggest ‘funeral,'” van Esch says. “It’s not wrong, per se. Maybe this is more common than ‘my Grandma’s rave party.’ But at the same time, it’s not something that you want to be reminded about. So it’s better to be a bit careful.”

Users seem to prefer this discretion. Keyboards get roundly criticized when a sexual, morbid, or otherwise disturbing phrase does get predicted. It’s likely that a lot more filtering happens behind the scenes before we even notice it. Janelle Shane, the creator of AI Weirdness, experiences lapses in machine judgment all the time. “Whenever I produce an AI experiment, I’m definitely filtering out offensive content, even when the training data is as innocuous as My Little Pony names. There’s no text-generating algorithm I would trust not to be offensive at some point.”

The true goal of text prediction can’t be as simple as anticipating what a user might want to type. After all, people often type things about sex or death—according to Google Ngrams, “job” is the most common noun after “blow,” and “bucket” is very common after “kick the.” But I experimentally typed these and similar taboo-but-common phrases into my phone’s keyboard, and it never predicted them straightaway. It waited until I’d typed most of the letters of the final word, until I’d definitely committed to the taboo, rather than reminding me of weighty topics when I wasn’t necessarily already thinking about them. With innocuous idioms (like “raining cats and”), the keyboard seemed more proactive about predicting them.

Instead, the goal of text prediction must be to anticipate what the user might want the machine to think they might want to type. For mundane topics, these two goals might seem identical, but their difference shows up as soon as a hint of controversy enters the picture. Predictive text needs to project an aspirational version of a user’s thoughts, a version that avoids subjects like sex and death even though these might be the most important topics to human existence—quite literally the way we enter and leave the world.

We prefer the keyboard to balance raw statistics against our feelings. Sex Death Phone Keyboard is a pretty good name for my future metal band (and a very bad name for my future pony), but I can’t say I’d actually buy a phone that reminds me of my own mortality when I’m composing a grocery list or suggests innuendos when I’m replying to a work email.

The predictive text meme is comforting in a social media world that often leaps from one dismal news cycle to the next. The customizations make us feel seen. The random quirks give our pattern-seeking brains delightful connections. The parts that don’t make sense reassure us of human superiority—the machines can’t be taking over yet if they can’t even write me a decent horoscope! And the topic boundaries prevent the meme from reminding us of our human frailty. The result is a version of ourselves through the verbal equivalent of an Instagram filter, eminently shareable on social media.

More Great WIRED Stories

Tech

[hostingnewsfeed]

Autocomplete Presents the Best Version of You

New Post has been published on http://webhostingtop3.com/autocomplete-presents-the-best-version-of-you/

Autocomplete Presents the Best Version of You

Type the phrase “In 2019, I’ll …” and let your smart­phone’s keyboard predict the rest. Depending on what else you’ve typed recently, you might end up with a result like one of these:

In 2019, I’ll let it be a surprise to be honest. In 2019, i’ll be alone. In 2019, I’ll be in the memes of the moment. In 2019, I’ll have to go to get the dog. In 2019 I will rule over the seven kingdoms or my name is not Aegon Targareon [sic].

Many variants on the predictive text meme—which works for both Android and iOS—can be found on social media. Not interested in predicting your 2019? Try writing your villain origin story by following your phone’s suggestions after typing “Foolish heroes! My true plan is …” Test the strength of your personal brand with “You should follow me on Twitter because …” Or launch your political career with “I am running for president with my running mate, @[3rd Twitter Suggestion], because we …”

Gretchen McCulloch is WIRED’s resident linguist. She’s the cocreator of Lingthusiasm, a podcast that’s enthusiastic about linguistics, and her book Because Internet: Understanding the New Rules of Language is coming out in July 2019 from Penguin.

In eight years, we’ve gone from Damn You Autocorrect to treating the strip of three predicted words as a sort of wacky but charming oracle. But when we try to practice divination by algorithm, we’re doing something more than killing a few minutes—we’re exploring the limits of what our devices can and cannot do.

Your phone’s keyboard comes with a basic list of words and sequences of words. That’s what powers the basic language features: autocorrect, where a sequence like “rhe” changes to “the” after you type it, and the suggestion strip just above the letters, which contains both completions (if you type “keyb” it might suggest “keyboard”) and next-word predictions (if you type “predictive” it might suggest “text,” “value,” and “analytics”). It’s this predictions feature that we use to generate amusing and slightly nonsensical strings of text—a function that goes beyond its intended purpose of supplying us with a word or two before we go back to tapping them out letter by letter.

The basic reason we get different results is that, as you use your phone, words or sequences of words that you type get added to your personal word list. “For most users, the on-device dictionary ends up containing local place-names, songs they like, and so on,” says Daan van Esch, a technical program manager of Gboard, Google’s keyboard for Android. Or, in the case of the “Aegon Targareon” example, slightly misspelled Game of Thrones characters.

Another factor that helps us get unique results is a slight bias toward predicting less frequent words. “Suggesting a very common word like ‘and’ might be less helpful because it’s short and easy to type,” van Esch says. “So maybe showing a longer word is actually more useful, even if it’s less frequent.” Of course, a longer word is probably going to be more interesting as meme fodder.

Finally, phones seem to choose different paths from the very beginning. Why are some people getting “I’ll be” while others get “I’ll have” or “I’ll let”? That part is probably not very exciting: The default Android keyboard presumably has slightly different predictions than the default iPhone keyboard, and third-party apps would also have slightly different predictions.

Whatever their provenance, the random juxtaposition of predictive text memes has become fodder for a growing genre of AI humor. Botnik Studios writes goofy songs using souped-up predictive keyboards and a lot of human tweaking. The blog AI Weirdness trains neural nets to do all sorts of ridiculous tasks, such as deciding whether a string of words is more likely to be a name from My Little Pony or a metal band. Darth Vader? 19 percent metal, 81 percent pony. Leia Organa? 96 percent metal, 4 percent pony. (I’m suddenly interpreting Star Wars in quite a new light.)

The combination of the customization and the randomness of the predictive text meme is compelling the way a BuzzFeed quiz or a horoscope is compelling—it gives you a tiny amount of insight into yourself to share, but not so much that you’re baring your soul. It’s also hard to get a truly terrible answer. In both cases, that’s by design.

You know how when you get a new phone and you have to teach it that, no, you aren’t trying to type “duck” and “ducking” all the time? Your keyboard deliberately errs on the conservative side. There are certain words that it just won’t try to complete, even if you get really close. After all, it’s better to accidentally send the word “public” when you meant “pubic” than the other way around.

This goes for sequences of words as well. Just because a sequence is common doesn’t mean it’s a good idea to predict it. “For a while, when you typed ‘I’m going to my Grandma’s,’ GBoard would actually suggest ‘funeral,'” van Esch says. “It’s not wrong, per se. Maybe this is more common than ‘my Grandma’s rave party.’ But at the same time, it’s not something that you want to be reminded about. So it’s better to be a bit careful.”

Users seem to prefer this discretion. Keyboards get roundly criticized when a sexual, morbid, or otherwise disturbing phrase does get predicted. It’s likely that a lot more filtering happens behind the scenes before we even notice it. Janelle Shane, the creator of AI Weirdness, experiences lapses in machine judgment all the time. “Whenever I produce an AI experiment, I’m definitely filtering out offensive content, even when the training data is as innocuous as My Little Pony names. There’s no text-generating algorithm I would trust not to be offensive at some point.”

The true goal of text prediction can’t be as simple as anticipating what a user might want to type. After all, people often type things about sex or death—according to Google Ngrams, “job” is the most common noun after “blow,” and “bucket” is very common after “kick the.” But I experimentally typed these and similar taboo-but-common phrases into my phone’s keyboard, and it never predicted them straightaway. It waited until I’d typed most of the letters of the final word, until I’d definitely committed to the taboo, rather than reminding me of weighty topics when I wasn’t necessarily already thinking about them. With innocuous idioms (like “raining cats and”), the keyboard seemed more proactive about predicting them.

Instead, the goal of text prediction must be to anticipate what the user might want the machine to think they might want to type. For mundane topics, these two goals might seem identical, but their difference shows up as soon as a hint of controversy enters the picture. Predictive text needs to project an aspirational version of a user’s thoughts, a version that avoids subjects like sex and death even though these might be the most important topics to human existence—quite literally the way we enter and leave the world.

We prefer the keyboard to balance raw statistics against our feelings. Sex Death Phone Keyboard is a pretty good name for my future metal band (and a very bad name for my future pony), but I can’t say I’d actually buy a phone that reminds me of my own mortality when I’m composing a grocery list or suggests innuendos when I’m replying to a work email.

The predictive text meme is comforting in a social media world that often leaps from one dismal news cycle to the next. The customizations make us feel seen. The random quirks give our pattern-seeking brains delightful connections. The parts that don’t make sense reassure us of human superiority—the machines can’t be taking over yet if they can’t even write me a decent horoscope! And the topic boundaries prevent the meme from reminding us of our human frailty. The result is a version of ourselves through the verbal equivalent of an Instagram filter, eminently shareable on social media.

More Great WIRED Stories

Tech

[hostingnewsfeed]

Autocomplete Presents the Best Version of You

New Post has been published on http://webhostingtop3.com/autocomplete-presents-the-best-version-of-you/

Autocomplete Presents the Best Version of You

Type the phrase “In 2019, I’ll …” and let your smart­phone’s keyboard predict the rest. Depending on what else you’ve typed recently, you might end up with a result like one of these:

In 2019, I’ll let it be a surprise to be honest. In 2019, i’ll be alone. In 2019, I’ll be in the memes of the moment. In 2019, I’ll have to go to get the dog. In 2019 I will rule over the seven kingdoms or my name is not Aegon Targareon [sic].

Many variants on the predictive text meme—which works for both Android and iOS—can be found on social media. Not interested in predicting your 2019? Try writing your villain origin story by following your phone’s suggestions after typing “Foolish heroes! My true plan is …” Test the strength of your personal brand with “You should follow me on Twitter because …” Or launch your political career with “I am running for president with my running mate, @[3rd Twitter Suggestion], because we …”

Gretchen McCulloch is WIRED’s resident linguist. She’s the cocreator of Lingthusiasm, a podcast that’s enthusiastic about linguistics, and her book Because Internet: Understanding the New Rules of Language is coming out in July 2019 from Penguin.

In eight years, we’ve gone from Damn You Autocorrect to treating the strip of three predicted words as a sort of wacky but charming oracle. But when we try to practice divination by algorithm, we’re doing something more than killing a few minutes—we’re exploring the limits of what our devices can and cannot do.

Your phone’s keyboard comes with a basic list of words and sequences of words. That’s what powers the basic language features: autocorrect, where a sequence like “rhe” changes to “the” after you type it, and the suggestion strip just above the letters, which contains both completions (if you type “keyb” it might suggest “keyboard”) and next-word predictions (if you type “predictive” it might suggest “text,” “value,” and “analytics”). It’s this predictions feature that we use to generate amusing and slightly nonsensical strings of text—a function that goes beyond its intended purpose of supplying us with a word or two before we go back to tapping them out letter by letter.

The basic reason we get different results is that, as you use your phone, words or sequences of words that you type get added to your personal word list. “For most users, the on-device dictionary ends up containing local place-names, songs they like, and so on,” says Daan van Esch, a technical program manager of Gboard, Google’s keyboard for Android. Or, in the case of the “Aegon Targareon” example, slightly misspelled Game of Thrones characters.

Another factor that helps us get unique results is a slight bias toward predicting less frequent words. “Suggesting a very common word like ‘and’ might be less helpful because it’s short and easy to type,” van Esch says. “So maybe showing a longer word is actually more useful, even if it’s less frequent.” Of course, a longer word is probably going to be more interesting as meme fodder.

Finally, phones seem to choose different paths from the very beginning. Why are some people getting “I’ll be” while others get “I’ll have” or “I’ll let”? That part is probably not very exciting: The default Android keyboard presumably has slightly different predictions than the default iPhone keyboard, and third-party apps would also have slightly different predictions.

Whatever their provenance, the random juxtaposition of predictive text memes has become fodder for a growing genre of AI humor. Botnik Studios writes goofy songs using souped-up predictive keyboards and a lot of human tweaking. The blog AI Weirdness trains neural nets to do all sorts of ridiculous tasks, such as deciding whether a string of words is more likely to be a name from My Little Pony or a metal band. Darth Vader? 19 percent metal, 81 percent pony. Leia Organa? 96 percent metal, 4 percent pony. (I’m suddenly interpreting Star Wars in quite a new light.)

The combination of the customization and the randomness of the predictive text meme is compelling the way a BuzzFeed quiz or a horoscope is compelling—it gives you a tiny amount of insight into yourself to share, but not so much that you’re baring your soul. It’s also hard to get a truly terrible answer. In both cases, that’s by design.

You know how when you get a new phone and you have to teach it that, no, you aren’t trying to type “duck” and “ducking” all the time? Your keyboard deliberately errs on the conservative side. There are certain words that it just won’t try to complete, even if you get really close. After all, it’s better to accidentally send the word “public” when you meant “pubic” than the other way around.

This goes for sequences of words as well. Just because a sequence is common doesn’t mean it’s a good idea to predict it. “For a while, when you typed ‘I’m going to my Grandma’s,’ GBoard would actually suggest ‘funeral,'” van Esch says. “It’s not wrong, per se. Maybe this is more common than ‘my Grandma’s rave party.’ But at the same time, it’s not something that you want to be reminded about. So it’s better to be a bit careful.”

Users seem to prefer this discretion. Keyboards get roundly criticized when a sexual, morbid, or otherwise disturbing phrase does get predicted. It’s likely that a lot more filtering happens behind the scenes before we even notice it. Janelle Shane, the creator of AI Weirdness, experiences lapses in machine judgment all the time. “Whenever I produce an AI experiment, I’m definitely filtering out offensive content, even when the training data is as innocuous as My Little Pony names. There’s no text-generating algorithm I would trust not to be offensive at some point.”

The true goal of text prediction can’t be as simple as anticipating what a user might want to type. After all, people often type things about sex or death—according to Google Ngrams, “job” is the most common noun after “blow,” and “bucket” is very common after “kick the.” But I experimentally typed these and similar taboo-but-common phrases into my phone’s keyboard, and it never predicted them straightaway. It waited until I’d typed most of the letters of the final word, until I’d definitely committed to the taboo, rather than reminding me of weighty topics when I wasn’t necessarily already thinking about them. With innocuous idioms (like “raining cats and”), the keyboard seemed more proactive about predicting them.

Instead, the goal of text prediction must be to anticipate what the user might want the machine to think they might want to type. For mundane topics, these two goals might seem identical, but their difference shows up as soon as a hint of controversy enters the picture. Predictive text needs to project an aspirational version of a user’s thoughts, a version that avoids subjects like sex and death even though these might be the most important topics to human existence—quite literally the way we enter and leave the world.

We prefer the keyboard to balance raw statistics against our feelings. Sex Death Phone Keyboard is a pretty good name for my future metal band (and a very bad name for my future pony), but I can’t say I’d actually buy a phone that reminds me of my own mortality when I’m composing a grocery list or suggests innuendos when I’m replying to a work email.

The predictive text meme is comforting in a social media world that often leaps from one dismal news cycle to the next. The customizations make us feel seen. The random quirks give our pattern-seeking brains delightful connections. The parts that don’t make sense reassure us of human superiority—the machines can’t be taking over yet if they can’t even write me a decent horoscope! And the topic boundaries prevent the meme from reminding us of our human frailty. The result is a version of ourselves through the verbal equivalent of an Instagram filter, eminently shareable on social media.

More Great WIRED Stories

Tech