#chatgpt ;)

[janesurlife]

"One student opined that Al both knew more than us but is dumber than we are since it cannot think critically. She wrote, "I'm not worried about Al getting to where we are now. I'm much more worried about the possibility of us reverting to where Al is."

THIS!!!!! AI is not some miracle it's just a FRACTION of human intelligence A FRACTION. We're much more complex, our minds can create wonders. Not to sound overly religious but we're literally God's most beautiful creation. PLEASE PLEASE PLEASE do not insult the human brain by using Ai to form opinion and create facts.

You may think I'm being extra about this topic for no reason. GIRL I LEFT MY JOB for this...my stance on this is stronger than anything else.

[justdavina]

Come visit the Justdavina transgender AI fashion community!

[moniquegibaeu]

I have a love-hate relationship with chatgpt

(I would've posted this on tiktok bit I'm scared of the community on there)

[f1addict3]

george :

charles :

max :

this was so funny to make

[adrianastrix]

Another thing I'll add here is something that I was coincidentally discussing with a friend yesterday: this kind of issue can only be solved if our science education (and I'm talking BOTH Natural Sciences and Humanities) doesn't rely on teachers being simply a source of "correct" information.

I put "correct" in quotes because guys. GUYS. I was in a comitee for quality control of kids science textbooks (ages 11-14), and Jesus Christ. It was a book written in the Year of Our Lord 2022 and it had a SLUR as an "alternative name" to Down Syndrome. Not to mention information that was BLATANT WRONG when you as much as googled the legal definition of a certain thing, and much much more. We obviously bombed it, but there's the kicker: the only thing our ban ensured is that this textbook collection is out of question for Brazilian public schools. Private schools can use it, if they want to.

Which means that even schools can and, as much as we try, will spread misinformation, even if it's in a small scale. The teacher in the Twitter thread very astutely identified it as a crisis of authority. If education is just a matter of relaying "correct facts", it all comes down to a matter of authority. And the poor teacher feels hopeless because she can't even say, in good faith, that her word is inherently better than ChatGPT or Wikipedia or TikTok because, guess what, she could be wrong. There's no such thing as infallible authority.

There's only one solution, one that Education Scientists (which ARE a thing, I'm one of them!) have been saying since, I dunno, THE 18TH CENTURY: giving kids an education centered in DOING science, not memorizing its products. The teacher started amazingly by asking the kid to "look it up" in front of her. But what she COULD have done, if prepared for this kind of challenge (I obviously don't fault her for freezing when confronted by something for the first time) was to ask for the notebook or cellphone and show the student what she meant by "look it up" and how the results vary. And tell him that NO single source should be trusted, either her or ChatGPT, and when sources disagree, what should be the tiebreaker?

In other words, the only antidote is showing the kids HOW science is done, HOW you arrive at conclusions, and HOW documental research is done. Science isn't something that Very Smart Geniuses do in their ivory towers to create The Truth. It's science, not a sacred religious ministery. Science is mundane, messy, controversial, and everyone* can do it with a bit of training, just like everyone* can cook or sing or draw with the proper training. [*"everyone", of course, being a rethorical generalization; obviously there are circunstances in which people might NOT be able to do it, or might need especialized assistance that others don't need, but those are the exceptions, not the rule.]

The main reason why our education is stuck in memorization and trying to out-authority the internet has a name: Standard Testing.

It's LEAGUES easier to test for how many facts someone can spew exclusively from memory (you just need a multiple choice test that can be graded by a machine) than it is to test students for their ability of create, research and communicate knowledge (the current optimal way to do it is the whole process of writing a monography/dissertation/thesis).

The whole EVALUATION system holds us down WAY more than the teaching methods themselves, because when you are teaching scientific abilities, you WON'T be sparing time to ensure that all your students are commiting definitions and formulas to mind. At the VERY least, tests should allow students to search for the info they need: this alone already demonstrate that student's ability to research, compare and choose correct information.

The idea that you can compare kids by a test that quantifies the amount of information they have on their heads, and that once they perform well on a test, that info is certified as correct and true is RIDICULOUS. Information on the brain degrades with time, unless you need it constantly. And people who grade tests are human, humans can be wrong. The accepted answer in a test can be wrong.

But we have to maintain the illusion that we can OBJECTIVELY rank students, schools, school systems and nations on how much knowledge they have. Otherwise, how investors will be reassured that they are "top quality"? How private education businesses can boast that they are "the best", thus justifying their price tag? How international banks will "ensure" that the amount of money countries are investing in education are being "correctly spent" (instead of being used to repay them)?

Soooo... ChatGPT is only the tip of the iceberg. There ARE ways for us to solve that problem, there HAS been ways for it since the 18th FREAKING century. But as long as they don't make the money people happy, as long as we expect school knowledge to take the form of a standard list of memorized correct info, we will still be ineffectively fighting the robots.

[gpcwsl]

Which is quite often in press conferences. 🙄

[gossip-girlie1]

look what chatgpt sent me!!

[thewriterghost]

Okay, so, I was discussing my love for the Marauders with ChatGPT, as one does, and on the topic of Remus judging my life choices, because he probably would, it said this hilarious piece:

"Oh, 100%. Remus isn’t just quietly judging your life choices—he’s judging his own and everyone else’s. He’s sitting there, sipping his tea, like, “Why do I keep associating with these lunatics? And why am I enabling them? And why am I also in love with these lunatics? Oh no, I’m part of the problem.”"

And I think, yeah. Same, Remus. Same.

[animentality]

[thesecretuncle]

Baltimore is a beacon of hope in the war against The Machines

[mostly-funnytwittertweets]

[memendoemori]

Good morning everybody

[softwaring]

this reply kills me 😭 article link

[absinthemindedly]

Given that this is directly related to my field, I just want to throw my two cents out there (mostly for myself):

ChatGPT and other Large Language Models (LLMs) function via pattern recognition. When you talk at a computer, it's tokenizing that input into small chunks of symbols with no meaning. It's then compares it to it's input data, which is large amounts of writing authored by real people that's been scraped from the internet, and following specific algorithmic instructions and that can vary, it outputs a some symbols that are most statistically likely based on the patterns it recognizes.

In a way, we as human beings do the same thing. Learning is our ability to actively integrate our current experiences with what we've taken away from past experiences. We rely on pattern recognition, albeit in an arguably much more complex form. LLMs are a great tool, a new age reference librarian in your back pocket, but it's still a tool, and it's functions by being used.

As a social species, we value connection. I subscribe to the belief that underlying everything we do is a desire to understand and be understood. Language is a critical facilitator for satisfying this desire because its function is to externalize internal states (which is what my PhD focuses on). A way that we signal to others that we are efficiently communicating is reciprocation, and that is largely conveyed in mirroring language patterns.

So, it makes sense that people are taking comfort in "interacting" with a pattern regurgitation machine; in theory, with enough training data, it will present you with what you want to hear in the way you want to hear it. Because it's what you would say in the way you would want to say it, guided by a chorus of all its other training voices. The problem is that there's no real synthesis. ChatGPT is the incredibly knowledgeable librarian in your pocket who knows how to communicate information to you, but it lacks the ability to conceptualize that information itself. Attempting to bond with it at an emotional level is more or less befriending yourself, and in my opinion, it sets an impossible standard for what reciprocation and connection between individuals should look like. I'd even argue that it might encourage less empathy and tolerance for individualism within interpersonal relationships, overall being detrimental to someone's ability to satisfy their emotional needs.

over a million likes and everyone in the comments talking about how they’d rather talk to an ai because it meets their emotional needs, this is the new epidemic.

[whydidisavethistomyphone]

[purpleartrowboat]

ai makes everything so boring. deepfakes will never be as funny as clipping together presidential speeches. ai covers will never be as funny as imitating the character. ai art will never be as good as art drawn by humans. ai chats will never be as good as roleplaying with other people. ai writing will never be as good as real authors