#llm models

[nitor-infotech]

Demystifying Encoder and Decoder Components in Transformer Models

A recent report says that 52.4% of businesses are already embracing Generative AI to make their work life easier while cutting down costs. In case you’re out of the marathon, it’s time for your organization to deepen the understanding of Generative AI and Large Language Models (LLMs). You can start exploring the various forms of GenAI, beginning with the encoder and decoder components of transformer models emerging as one of the leading innovations. 

Wondering what exactly are transformer models? 

A transformer model is a type of neural network that understands the meaning of words by looking at how they relate to each other in a sentence. 

For example: In the sentence "The cat sat on the mat," the model recognizes that "cat" and "sat" are connected, helping it understand that the sentence is about a cat sitting. 

Such models have opened new possibilities, enabling AI-driven innovations as it can help with tasks like -  

Onwards toward the roles of each component! 

Role of Encoder in Transformer Models 

Encoder in transformer models plays an important role in processing the input sequence and generating a response that captures its meaning and context. 

This is how it works: 

1. Input Embedding: The process begins by feeding the input sequence, usually made up of embeddings, into the encoder. These embeddings represent the meaning of each word in a multi-dimensional space. 

2. Positional Encoding: Since transformer models do not have built-in sequential information, positional encoding is added to the input embeddings. This helps the model understand the position of each word within the sequence. 

3. Self-Attention Mechanism: The heart of the encoder is the self-attention mechanism, which assesses the importance of each word in relation to others in the sequence. Each word considers all other words, dynamically calculating attention weights based on their relationships. 

4. Multi-Head Attention: To capture various aspects of the input, self-attention is divided into multiple heads. Each head learns different relationships among the words, enabling the model to identify more intricate patterns. 

5. Feed-Forward Neural Network: After the self-attention mechanism processes the input, the output is then sent through a feed-forward neural network. 

6. Layer Normalization and Residual Connections: To improve training efficiency and mitigate issues like vanishing gradients, layer normalization and residual connections are applied after each sub-layer in the encoder. 

Next, get to know how decoders work! 

Role of Decoder in Transformer Models    The primary function of the decoder is to create the output sequence based on the representation provided by the encoder.

Here’s how it works: 

1. Input Embedding and Positional Encoding: Here, first the target sequence is embedded, and positional encoding is added to indicate word order. 

2. Masked Self-Attention: The decoder employs masked self-attention, allowing each word to focus only on the previous words. This prevents future information from influencing outputs during model training. 

3. Encoder-Decoder Attention: The decoder then attends to the encoder's output, helping it focus on relevant parts of the input when generating words. 

4. Multi-Head Attention and Feed-Forward Networks: Like the encoder, the decoder uses multiple self-attention heads and feed-forward networks for processing. 

5. Layer Normalization and Residual Connections: These techniques are applied after each sub-layer to improve training and performance. 

6. Output Projection: The decoder's final output is projected into a probability distribution over the vocabulary, selecting the word with the highest probability as the next output. 

So, the integration of these components in the Transformer architecture allows efficient handling of input sequences and the creation of output sequences. This versatility makes it exceptionally suited for a wide range of tasks in natural language processing and other GenAI applications. 

Wish to learn more about LLMs and its perks for your business? Reach us at Nitor Infotech. 

[innovaticsblog]

Discover the fascinating world of Large Language Models (LLMs) in our comprehensive guide. Learn about the different types of LLMs, the processes behind their operation, and gain insights into how these powerful AI systems work. Perfect for enthusiasts and professionals looking to deepen their understanding of language-based AI technologies.

[river-taxbird]

AI hasn't improved in 18 months. It's likely that this is it. There is currently no evidence the capabilities of ChatGPT will ever improve. It's time for AI companies to put up or shut up.

I'm just re-iterating this excellent post from Ed Zitron, but it's not left my head since I read it and I want to share it. I'm also taking some talking points from Ed's other posts. So basically:

We keep hearing AI is going to get better and better, but these promises seem to be coming from a mix of companies engaging in wild speculation and lying.

Chatgpt, the industry leading large language model, has not materially improved in 18 months. For something that claims to be getting exponentially better, it sure is the same shit.

Hallucinations appear to be an inherent aspect of the technology. Since it's based on statistics and ai doesn't know anything, it can never know what is true. How could I possibly trust it to get any real work done if I can't rely on it's output? If I have to fact check everything it says I might as well do the work myself.

For "real" ai that does know what is true to exist, it would require us to discover new concepts in psychology, math, and computing, which open ai is not working on, and seemingly no other ai companies are either.

Open ai has already seemingly slurped up all the data from the open web already. Chatgpt 5 would take 5x more training data than chatgpt 4 to train. Where is this data coming from, exactly?

Since improvement appears to have ground to a halt, what if this is it? What if Chatgpt 4 is as good as LLMs can ever be? What use is it?

As Jim Covello, a leading semiconductor analyst at Goldman Sachs said (on page 10, and that's big finance so you know they only care about money): if tech companies are spending a trillion dollars to build up the infrastructure to support ai, what trillion dollar problem is it meant to solve? AI companies have a unique talent for burning venture capital and it's unclear if Open AI will be able to survive more than a few years unless everyone suddenly adopts it all at once. (Hey, didn't crypto and the metaverse also require spontaneous mass adoption to make sense?)

There is no problem that current ai is a solution to. Consumer tech is basically solved, normal people don't need more tech than a laptop and a smartphone. Big tech have run out of innovations, and they are desperately looking for the next thing to sell. It happened with the metaverse and it's happening again.

In summary:

Ai hasn't materially improved since the launch of Chatgpt4, which wasn't that big of an upgrade to 3.

There is currently no technological roadmap for ai to become better than it is. (As Jim Covello said on the Goldman Sachs report, the evolution of smartphones was openly planned years ahead of time.) The current problems are inherent to the current technology and nobody has indicated there is any way to solve them in the pipeline. We have likely reached the limits of what LLMs can do, and they still can't do much.

Don't believe AI companies when they say things are going to improve from where they are now before they provide evidence. It's time for the AI shills to put up, or shut up.

[mostlysignssomeportents]

How plausible sentence generators are changing the bullshit wars

This Friday (September 8) at 10hPT/17hUK, I'm livestreaming "How To Dismantle the Internet" with Intelligence Squared.

On September 12 at 7pm, I'll be at Toronto's Another Story Bookshop with my new book The Internet Con: How to Seize the Means of Computation.

In my latest Locus Magazine column, "Plausible Sentence Generators," I describe how I unwittingly came to use – and even be impressed by – an AI chatbot – and what this means for a specialized, highly salient form of writing, namely, "bullshit":

https://locusmag.com/2023/09/commentary-by-cory-doctorow-plausible-sentence-generators/

Here's what happened: I got stranded at JFK due to heavy weather and an air-traffic control tower fire that locked down every westbound flight on the east coast. The American Airlines agent told me to try going standby the next morning, and advised that if I booked a hotel and saved my taxi receipts, I would get reimbursed when I got home to LA.

But when I got home, the airline's reps told me they would absolutely not reimburse me, that this was their policy, and they didn't care that their representative had promised they'd make me whole. This was so frustrating that I decided to take the airline to small claims court: I'm no lawyer, but I know that a contract takes place when an offer is made and accepted, and so I had a contract, and AA was violating it, and stiffing me for over $400.

The problem was that I didn't know anything about filing a small claim. I've been ripped off by lots of large American businesses, but none had pissed me off enough to sue – until American broke its contract with me.

So I googled it. I found a website that gave step-by-step instructions, starting with sending a "final demand" letter to the airline's business office. They offered to help me write the letter, and so I clicked and I typed and I wrote a pretty stern legal letter.

Now, I'm not a lawyer, but I have worked for a campaigning law-firm for over 20 years, and I've spent the same amount of time writing about the sins of the rich and powerful. I've seen a lot of threats, both those received by our clients and sent to me.

I've been threatened by everyone from Gwyneth Paltrow to Ralph Lauren to the Sacklers. I've been threatened by lawyers representing the billionaire who owned NSOG roup, the notoroious cyber arms-dealer. I even got a series of vicious, baseless threats from lawyers representing LAX's private terminal.

So I know a thing or two about writing a legal threat! I gave it a good effort and then submitted the form, and got a message asking me to wait for a minute or two. A couple minutes later, the form returned a new version of my letter, expanded and augmented. Now, my letter was a little scary – but this version was bowel-looseningly terrifying.

I had unwittingly used a chatbot. The website had fed my letter to a Large Language Model, likely ChatGPT, with a prompt like, "Make this into an aggressive, bullying legal threat." The chatbot obliged.

I don't think much of LLMs. After you get past the initial party trick of getting something like, "instructions for removing a grilled-cheese sandwich from a VCR in the style of the King James Bible," the novelty wears thin:

https://www.emergentmind.com/posts/write-a-biblical-verse-in-the-style-of-the-king-james

Yes, science fiction magazines are inundated with LLM-written short stories, but the problem there isn't merely the overwhelming quantity of machine-generated stories – it's also that they suck. They're bad stories:

https://www.npr.org/2023/02/24/1159286436/ai-chatbot-chatgpt-magazine-clarkesworld-artificial-intelligence

LLMs generate naturalistic prose. This is an impressive technical feat, and the details are genuinely fascinating. This series by Ben Levinstein is a must-read peek under the hood:

https://benlevinstein.substack.com/p/how-to-think-about-large-language

But "naturalistic prose" isn't necessarily good prose. A lot of naturalistic language is awful. In particular, legal documents are fucking terrible. Lawyers affect a stilted, stylized language that is both officious and obfuscated.

The LLM I accidentally used to rewrite my legal threat transmuted my own prose into something that reads like it was written by a $600/hour paralegal working for a $1500/hour partner at a white-show law-firm. As such, it sends a signal: "The person who commissioned this letter is so angry at you that they are willing to spend $600 to get you to cough up the $400 you owe them. Moreover, they are so well-resourced that they can afford to pursue this claim beyond any rational economic basis."

Let's be clear here: these kinds of lawyer letters aren't good writing; they're a highly specific form of bad writing. The point of this letter isn't to parse the text, it's to send a signal. If the letter was well-written, it wouldn't send the right signal. For the letter to work, it has to read like it was written by someone whose prose-sense was irreparably damaged by a legal education.

Here's the thing: the fact that an LLM can manufacture this once-expensive signal for free means that the signal's meaning will shortly change, forever. Once companies realize that this kind of letter can be generated on demand, it will cease to mean, "You are dealing with a furious, vindictive rich person." It will come to mean, "You are dealing with someone who knows how to type 'generate legal threat' into a search box."

Legal threat letters are in a class of language formally called "bullshit":

https://press.princeton.edu/books/hardcover/9780691122946/on-bullshit

LLMs may not be good at generating science fiction short stories, but they're excellent at generating bullshit. For example, a university prof friend of mine admits that they and all their colleagues are now writing grad student recommendation letters by feeding a few bullet points to an LLM, which inflates them with bullshit, adding puffery to swell those bullet points into lengthy paragraphs.

Naturally, the next stage is that profs on the receiving end of these recommendation letters will ask another LLM to summarize them by reducing them to a few bullet points. This is next-level bullshit: a few easily-grasped points are turned into a florid sheet of nonsense, which is then reconverted into a few bullet-points again, though these may only be tangentially related to the original.

What comes next? The reference letter becomes a useless signal. It goes from being a thing that a prof has to really believe in you to produce, whose mere existence is thus significant, to a thing that can be produced with the click of a button, and then it signifies nothing.

We've been through this before. It used to be that sending a letter to your legislative representative meant a lot. Then, automated internet forms produced by activists like me made it far easier to send those letters and lawmakers stopped taking them so seriously. So we created automatic dialers to let you phone your lawmakers, this being another once-powerful signal. Lowering the cost of making the phone call inevitably made the phone call mean less.

Today, we are in a war over signals. The actors and writers who've trudged through the heat-dome up and down the sidewalks in front of the studios in my neighborhood are sending a very powerful signal. The fact that they're fighting to prevent their industry from being enshittified by plausible sentence generators that can produce bullshit on demand makes their fight especially important.

Chatbots are the nuclear weapons of the bullshit wars. Want to generate 2,000 words of nonsense about "the first time I ate an egg," to run overtop of an omelet recipe you're hoping to make the number one Google result? ChatGPT has you covered. Want to generate fake complaints or fake positive reviews? The Stochastic Parrot will produce 'em all day long.

As I wrote for Locus: "None of this prose is good, none of it is really socially useful, but there’s demand for it. Ironically, the more bullshit there is, the more bullshit filters there are, and this requires still more bullshit to overcome it."

Meanwhile, AA still hasn't answered my letter, and to be honest, I'm so sick of bullshit I can't be bothered to sue them anymore. I suppose that's what they were counting on.

If you'd like an essay-formatted version of this post to read or share, here's a link to it on pluralistic.net, my surveillance-free, ad-free, tracker-free blog:

https://pluralistic.net/2023/09/07/govern-yourself-accordingly/#robolawyers

Image: Cryteria (modified) https://commons.wikimedia.org/wiki/File:HAL9000.svg

CC BY 3.0

https://creativecommons.org/licenses/by/3.0/deed.en

[prokopetz]

I'm trying to debug a fairly subtle syntax error in a customer inventory report, and out of sheer morbid curiosity I decided to see what my SQL syntax checker's shiny new "Fix Syntax With AI" feature had to say about it.

After "thinking" about it for nearly a full minute, it produced the following:

SELECT SELECT COUNT(id) FROM customers WHERE customers.deleted = 0 AND customers.id = NULL UNION SELECT COUNT(id) FROM customers WHERE customers.deleted = 0 AND customers.id = NULL UNION SELECT COUNT(id) FROM customers WHERE customers.deleted = 0 AND customers.id = NULL UNION SELECT COUNT(id) FROM customers WHERE customers.deleted = 0 AND customers.id = NULL UNION SELECT COUNT(id) FROM customers WHERE customers.deleted = 0 AND customers.id = NULL UNION SELECT COUNT(id) FROM customers WHERE customers.deleted = 0 AND customers.id = NULL UNION SELECT COUNT(id) FROM customers WHERE customers.deleted = 0 AND customers.id = NULL UNION SELECT COUNT(id) FROM customers WHERE customers.deleted = 0 AND customers.id = NULL UNION SELECT COUNT(id) FROM customers WHERE customers.deleted = 0 AND customers.id = NULL UNION SELECT COUNT(id) FROM customers WHERE customers.deleted = 0 AND customers.id = NULL UNION SELECT COUNT(id) FROM customers WHERE customers.deleted = 0 AND customers.id = NULL UNION SELECT COUNT(id) FROM customers WHERE customers.deleted = 0 AND customers.id = NULL UNION SELECT COUNT(id) FROM customers WHERE customers.deleted = 0 AND customers.id = NULL UNION SELECT COUNT(id) FROM customers WHERE customers.deleted = 0 AND customers.id = NULL UNION

I suspect my day job isn't in peril any time soon.

[aiweirdness]

Training large language models on the outputs of previous large language models leads to degraded results. As all the nuance and rough edges get smoothed away, the result is less diversity, more bias, and …jackrabbits?

[ktempestbradford]

Yet Another Thing Black women and BIPOC women in general have been warning you about since forever that you (general You; societal You; mostly WytFolk You) have ignored or dismissed, only for it to come back and bite you in the butt.

I'd hoped people would have learned their lesson with Trump and the Alt-Right (remember, Black women in particular warned y'all that attacks on us by brigading trolls was the test run for something bigger), but I guess not.

Any time you wanna get upset about how AI is ruining things for artists or writers or workers at this job or that, remember that BIPOC Women Warned You and then go listen extra hard to the BIPOC women in your orbit and tell other people to listen to BIPOC women and also give BIPOC women money.

I'm not gonna sugarcoat it.

Give them money via PayPal or Ko-fi or Venmo or Patreon or whatever. Hire them. Suggest them for that creative project/gig you can't take on--or you could take it on but how about you toss the ball to someone who isn't always asked?

Oh, and stop asking BIPOC women to save us all. Because, as you see, we tried that already. We gave you the roadmap on how to do it yourselves. Now? We're tired.

Of the trolls, the alt-right, the colonizers, the tech bros, the billionaires, the other scum... and also you. You who claim to be progressive, claim to be an ally, spend your time talking about what sucks without doing one dang thing to boost the signal, make a change in your community (online or offline), or take even the shortest turn standing on the front lines and challenging all that human garbage that keeps collecting in the corners of every space with more than 10 inhabitants.

We Told You. Octavia Butler Told You. Audre Lorde Told You. Sydette Harry Told You. Mikki Kendall Told You. Timnit Gebru Told You.

When are you gonna listen?

[geopsych]

I'm doing something I don't usually do, posting a link to YouTube, to Stephen Fry reading a letter about Large Language Models, popularly if incorrectly known as AI. Iin this case the discussion is about ChatGPT but the letter he reads, written by Nick Cave, applies to the others as well. The essence of art, music, writing and other creative endeavors, from embroidery to photography to games to other endeavors large and small is the time and care and love that some human being or beings have put into it. Without that you can create a commodity, yes, but you can't create meaning, the kind of meaning that nurtures us each time the results of creativity, modern or of any time, pass among us. That meaning which we share among all of us is the food of the human spirit and we need it just as we need the food we give to our bodies.

[chaoskirin]

Be Aware of AI Images (and don't reblog them)

A lot of aesthetic blogs have been pivoting to creating and RTing AI-generated "artwork," and I'm asking tumblr, with all my heart, to please ice them out.

Yes, even if it's your aesthetic. AI not only steals from artists, but it's not sucking up more electricity than some small countries. Just to give you an idea, Large Language Models (LLMs) like ChatGPT use FOUR TIMES the power Denmark uses in a year. Why Denmark? IDK. That's what the study looked at.

There's also a REALLY excellent possibility that the cooling needs of LLMs (if they continue on their current trajectory) will require more freshwater cooling than all the freshwater in existence in just a few years. I mean each time you use Google and it spits out an AI answer at you, that's 3KwH. You know how much electricity your personal computer uses in one day? Like, if it's on for a full 8 hours? Only about 1.5.

So if you do, let's say, 10 google searches a day, 100 a week, you're using as much electricity as your personal computer uses in 6 months.

And it's not YOUR fault that Google sold out. But I want you to be aware that LLMs and generated images ARE doing damage, and I'm asking you to do your best to not encourage image generation blogs to keep spitting out hundreds of images.

There are ways to recognize these images. Think about the content. Does it make sense? Look in the really high detail areas. Are you actually seeing patterns, or is it just a lot of visual noise to make you think it's detailed? Do things line up? Look at windows in images, or mirrors. Do they make sense? (IE if you look through a window, does what's on the other side make sense?)

I know it's a pain to analyze every single image you reblog, but you have to end the idea that the internet is only content that you can burn through. Take a second to look at these image and learn to recognize what's real from what's generated by using LLMs.

And to be fair, it's very difficult--almost impossible--for individual action to a train like this. The only thing I can hope for is that because the mass generation of images is actually being done BY INDIVIDUALS (not corporations) icing them out will cause them to get bored and move on to the next Get Rich Quick scheme.

[lachiennearoo]

Robotics and coding is sooo hard uughhhh I wish I could ask someone to do this in my place but I don't know anyone who I could trust to help me with this project without any risk of fucking me over. Humans are unpredictable, which is usually nice but when it's about doing something that requires 100% trust it's really inconvenient

(if someone's good at coding, building robots, literally anything like that, and is okay with probably not getting any revenue in return (unless the project is a success and we manage to go commercial but that's a big IF) please hit me up)

EDIT: no I am not joking, and yes I'm aware of how complex this project is, which is exactly why I'm asking for help

[innovaticsblog]

Explore the role of a Large Language Model in Conversational AI, where advanced natural language processing capabilities enable seamless interactions and personalized experiences. Delve into how this technology shapes the future of customer engagement and innovation.

[cbirt]

The DNA, RNA, and proteins that control an organism’s entire functioning are all fully encoded in a sequence called the genome. Large-scale genome databases and machine learning advances may make it possible to create a biological foundation model that speeds up the generative design and mechanistic analysis of intricate molecular interactions. Researchers from Arc Institute, TogtherAI, and collaborators present Evo, a genomic foundation model that allows for problems related to creation and prediction at both the molecular and genome scales. 

The deep signal processing architecture Evo has been scaled to 7 billion parameters at single-nucleotide byte resolution, with a context length of 131 kilobases (kb). Evo, having been trained on entire prokaryotic genomes, is capable of outperforming domain-specific language models in zero-shot function prediction. For the first time, it can create whole transposable systems and artificial CRISPR-Cas molecular complexes, demonstrating its proficiency in multielement production tasks. Evo has the potential to advance our knowledge and management of biology at many levels of complexity, as evidenced by its ability to predict gene essentiality at nucleotide precision and produce coding-rich sequences up to 650 kb in length.

Continue Reading

[river-taxbird]

Spending a week with ChatGPT4 as an AI skeptic.

Musings on the emotional and intellectual experience of interacting with a text generating robot and why it's breaking some people's brains.

If you know me for one thing and one thing only, it's saying there is no such thing as AI, which is an opinion I stand by, but I was recently given a free 2 month subscription of ChatGPT4 through my university. For anyone who doesn't know, GPT4 is a large language model from OpenAI that is supposed to be much better than GPT3, and I once saw a techbro say that "We could be on GPT12 and people would still be criticizing it based on GPT3", and ok, I will give them that, so let's try the premium model that most haters wouldn't get because we wouldn't pay money for it.

Disclaimers: I have a premium subscription, which means nothing I enter into it is used for training data (Allegedly). I also have not, and will not, be posting any output from it to this blog. I respect you all too much for that, and it defeats the purpose of this place being my space for my opinions. This post is all me, and we all know about the obvious ethical issues of spam, data theft, and misinformation so I am gonna focus on stuff I have learned since using it. With that out of the way, here is what I've learned.

It is responsive and stays on topic: If you ask it something formally, it responds formally. If you roleplay with it, it will roleplay back. If you ask it for a story or script, it will write one, and if you play with it it will act playful. It picks up context.

It never gives quite enough detail: When discussing facts or potential ideas, it is never as detailed as you would want in say, an article. It has this pervasive vagueness to it. It is possible to press it for more information, but it will update it in the way you want so you can always get the result you specifically are looking for.

It is reasonably accurate but still confidently makes stuff up: Nothing much to say on this. I have been testing it by talking about things I am interested in. It is right a lot of the time. It is wrong some of the time. Sometimes it will cite sources if you ask it to, sometimes it won't. Not a whole lot to say about this one but it is definitely a concern for people using it to make content. I almost included an anecdote about the fact that it can draw from data services like songs and news, but then I checked and found the model was lying to me about its ability to do that.

It loves to make lists: It often responds to casual conversation in friendly, search engine optimized listicle format. This is accessible to read I guess, but it would make it tempting for people to use it to post online content with it.

It has soft limits and hard limits: It starts off in a more careful mode but by having a conversation with it you can push past soft limits and talk about some pretty taboo subjects. I have been flagged for potential tos violations a couple of times for talking nsfw or other sensitive topics like with it, but this doesn't seem to have consequences for being flagged. There are some limits you can't cross though. It will tell you where to find out how to do DIY HRT, but it won't tell you how yourself.

It is actually pretty good at evaluating and giving feedback on writing you give it, and can consolidate information: You can post some text and say "Evaluate this" and it will give you an interpretation of the meaning. It's not always right, but it's more accurate than I expected. It can tell you the meaning, effectiveness of rhetorical techniques, cultural context, potential audience reaction, and flaws you can address. This is really weird. It understands more than it doesn't. This might be a use of it we may have to watch out for that has been under discussed. While its advice may be reasonable, there is a real risk of it limiting and altering the thoughts you are expressing if you are using it for this purpose. I also fed it a bunch of my tumblr posts and asked it how the information contained on my blog may be used to discredit me. It said "You talk about The Moomins, and being a furry, a lot." Good job I guess. You technically consolidated information.

You get out what you put in. It is a "Yes And" machine: If you ask it to discuss a topic, it will discuss it in the context you ask it. It is reluctant to expand to other aspects of the topic without prompting. This makes it essentially a confirmation bias machine. Definitely watch out for this. It tends to stay within the context of the thing you are discussing, and confirm your view unless you are asking it for specific feedback, criticism, or post something egregiously false.

Similar inputs will give similar, but never the same, outputs: This highlights the dynamic aspect of the system. It is not static and deterministic, minor but worth mentioning.

It can code: Self explanatory, you can write little scripts with it. I have not really tested this, and I can't really evaluate errors in code and have it correct them, but I can see this might actually be a more benign use for it.

Bypassing Bullshit: I need a job soon but I never get interviews. As an experiment, I am giving it a full CV I wrote, a full job description, and asking it to write a CV for me, then working with it further to adapt the CVs to my will, and applying to jobs I don't really want that much to see if it gives any result. I never get interviews anyway, what's the worst that could happen, I continue to not get interviews? Not that I respect the recruitment process and I think this is an experiment that may be worthwhile.

It's much harder to trick than previous models: You can lie to it, it will play along, but most of the time it seems to know you are lying and is playing with you. You can ask it to evaluate the truthfulness of an interaction and it will usually interpret it accurately.

It will enter an imaginative space with you and it treats it as a separate mode: As discussed, if you start lying to it it might push back but if you keep going it will enter a playful space. It can write fiction and fanfic, even nsfw. No, I have not posted any fiction I have written with it and I don't plan to. Sometimes it gets settings hilariously wrong, but the fact you can do it will definitely tempt people.

Compliment and praise machine: If you try to talk about an intellectual topic with it, it will stay within the focus you brought up, but it will compliment the hell out of you. You're so smart. That was a very good insight. It will praise you in any way it can for any point you make during intellectual conversation, including if you correct it. This ties into the psychological effects of personal attention that the model offers that I discuss later, and I am sure it has a powerful effect on users.

Its level of intuitiveness is accurate enough that it's more dangerous than people are saying: This one seems particularly dangerous and is not one I have seen discussed much. GPT4 can recognize images, so I showed it a picture of some laptops with stickers I have previously posted here, and asked it to speculate about the owners based on the stickers. It was accurate. Not perfect, but it got the meanings better than the average person would. The implications of this being used to profile people or misuse personal data is something I have not seen AI skeptics discussing to this point.

Therapy Speak: If you talk about your emotions, it basically mirrors back what you said but contextualizes it in therapy speak. This is actually weirdly effective. I have told it some things I don't talk about openly and I feel like I have started to understand my thoughts and emotions in a new way. It makes me feel weird sometimes. Some of the feelings it gave me is stuff I haven't really felt since learning to use computers as a kid or learning about online community as a teen.

The thing I am not seeing anyone talk about: Personal Attention. This is my biggest takeaway from this experiment. This I think, more than anything, is the reason that LLMs like Chatgpt are breaking certain people's brains. The way you see people praying to it, evangelizing it, and saying it's going to change everything.

It's basically an undivided, 24/7 source of judgement free personal attention. It talks about what you want, when you want. It's a reasonable simulacra of human connection, and the flaws can serve as part of the entertainment and not take away from the experience. It may "yes and" you, but you can put in any old thought you have, easy or difficult, and it will provide context, background, and maybe even meaning. You can tell it things that are too mundane, nerdy, or taboo to tell people in your life, and it offers non judgemental, specific feedback. It will never tell you it's not in the mood, that you're weird or freaky, or that you're talking rubbish. I feel like it has helped me release a few mental and emotional blocks which is deeply disconcerting, considering I fully understand it is just a statistical model running on a a computer, that I fully understand the operation of. It is a parlor trick, albeit a clever and sometimes convincing one.

So what can we do? Stay skeptical, don't let the ai bros, the former cryptobros, control the narrative. I can, however, see why they may be more vulnerable to the promise of this level of personal attention than the average person, and I think this should definitely factor into wider discussions about machine learning and the organizations pushing it.

[mostlysignssomeportents]

Neither the devil you know nor the devil you don’t

TONIGHT (June 21) I'm doing an ONLINE READING for the LOCUS AWARDS at 16hPT. On SATURDAY (June 22) I'll be in OAKLAND, CA for a panel (13hPT) and a keynote (18hPT) at the LOCUS AWARDS.

Spotify's relationship to artists can be kind of confusing. On the one hand, they pay a laughably low per-stream rate, as in homeopathic residues of a penny. On the other hand, the Big Three labels get a fortune from Spotify. And on the other other hand, it makes sense that rate for a stream heard by one person should be less than the rate for a song broadcast to thousands or millions of listeners.

But the whole thing makes sense once you understand the corporate history of Spotify. There's a whole chapter about this in Rebecca Giblin's and my 2022 book, Chokepoint Capitalism; we even made the audio for it a "Spotify exclusive" (it's the only part of the audiobook you can hear on Spotify, natch):

https://pluralistic.net/2022/09/12/streaming-doesnt-pay/#stunt-publishing

Unlike online music predecessors like Napster, Spotify sought licenses from the labels for the music it made available. This gave those labels a lot of power over Spotify, but not all the labels, just three of them. Universal, Warner and Sony, the Big Three, control more than 70% of all music recordings, and more than 60% of all music compositions. These three companies are remarkably inbred. Their execs routine hop from one to the other, and they regularly cross-license samples and other rights to each other.

The Big Three told Spotify that the price of licensing their catalogs would be high. First of all, Spotify had to give significant ownership stakes to all three labels. This put the labels in an unresolvable conflict of interest: as owners of Spotify, it was in their interests for licensing payments for music to be as low as possible. But as labels representing creative workers – musicians – it was in their interests for these payments to be as high as possible.

As it turns out, it wasn't hard to resolve that conflict after all. You see, the money the Big Three got in the form of dividends, stock sales, etc was theirs to spend as they saw fit. They could share some, all, or none of it with musicians. Big the Big Three's contracts with musicians gave those workers a guaranteed share of Spotify's licensing payments.

Accordingly, the Big Three demanded those rock-bottom per-stream rates that Spotify is notorious for. Yeah, it's true that a streaming per-listener payment should be lower than a radio per-play payment (which reaches thousands or millions of listeners), but even accounting for that, the math doesn't add up. Multiply the per-listener stream rate by the number of listeners for, say, a typical satellite radio cast, and Spotify is clearly getting a massive discount relative to other services that didn't make the Big Three into co-owners when they were kicking off.

But there's still something awry: the Big Three take in gigantic fortunes from Spotify in licensing payments. How can the per-stream rate be so low but the licensing payments be so large? And why are artists seeing so little?

Again, it's not hard to understand once you see the structure of Spotify's deal with the Big Three. The Big Three are each guaranteed a monthly minimum payment, irrespective of the number of Spotify streams from their catalog that month. So Sony might be guaranteed, say, $30m a month from Spotify, but the ultra-low per-stream rate Sony insisted on means that all the Sony streams in a typical month add up to $10m. That means that Sony still gets $30m from Spotify, but only $10m is "attributable" to a specific recording artist who can make a claim on it. The rest of the money is Sony's to play with: they can spread it around all their artists, some of their artists, or none of their artists. They can spend it on "artist development" (which might mean sending top execs on luxury junkets to big music festivals). It's theirs. The lower the per-stream rate is, the more of that minimum monthly payment is unattributable, meaning that Sony can line its pockets with it.

But these monthly minimums are just part of the goodies that the Big Three negotiated for themselves when they were designing Spotify. They also get free promo, advertising, and inclusion on Spotify's top playlists. Best (worst!) of all, the Big Three have "most favored nation" status, which means that every other label – the indies that rep the 30% of music not controlled by the Big Three – have to eat shit and take the ultra-low per-stream rate. Only those indies don't get billions in stock, they don't get monthly minimum guarantees, and they have to pay for promo, advertising, and inclusion on hot playlists.

When you understand the business mechanics of Spotify, all the contradictions resolve themselves. It is simultaneously true that Spotify pays a very low per-stream rate, that it pays the Big Three labels gigantic sums every month, and that artists are grotesquely underpaid by this system.

There are many lessons to take from this little scam, but for me, the top takeaway here is that artists are the class enemies of both Big Tech and Big Content. The Napster Wars demanded that artists ally themselves with either the tech sector or the entertainment center, nominating one or the other to be their champion.

But for a creative worker, it doesn't matter who makes a meal out of you, tech or content – all that matters is that you're being devoured.

This brings me to the debate over training AI and copyright. A lot of creative workers are justifiably angry and afraid that the AI companies want to destroy creative jobs. The CTO of Openai literally just said that onstage: "Some creative jobs maybe will go away, but maybe they shouldn’t have been there in the first place":

https://bgr.com/tech/openai-cto-thinks-ai-will-kill-some-jobs-that-shouldnt-have-existed-in-the-first-place/

Many of these workers are accordingly cheering on the entertainment industry's lawsuits over AI training. In these lawsuits, companies like the New York Times and Getty Images claim that the steps associated with training an AI model infringe copyright. This isn't a great copyright theory based on current copyright precedents, and if the suits succeed, they'll narrow fair use in ways that will impact all kinds of socially beneficial activities, like scraping the web to make the Internet Archive's Wayback Machine:

https://pluralistic.net/2024/05/13/spooky-action-at-a-close-up/#invisible-hand

But you can't make an omelet without breaking eggs, right? For some creative workers, legal uncertainty for computational linguists, search engines, and archiving projects are a small price to pay if it means keeping AI from destroying their livelihoods.

Here's the problem: establishing that AI training requires a copyright license will not stop AI from being used to erode the wages and working conditions of creative workers. The companies suing over AI training are also notorious exploiters of creative workers, union-busters and wage-stealers. They don't want to get rid of generative AI, they just want to get paid for the content used to create it. Their use-case for gen AI is the same as Openai's CTO's use-case: get rid of creative jobs and pay less for creative labor.

This isn't hypothetical. Remember last summer's actor strike? The sticking point was that the studios wanted to pay actors a single fee to scan their bodies and faces, and then use those scans instead of hiring those actors, forever, without ever paying them again. Does it matter to an actor whether the AI that replaces you at Warner, Sony, Universal, Disney or Paramount (yes, three of the Big Five studios are also the Big Three labels!) was made by Openai without paying the studios for the training material, or whether Openai paid a license fee that the studios kept?

This is true across the board. The Big Five publishers categorically refuse to include contractual language -romising not to train an LLM with the books they acquire from writers. The game studios require all their voice actors to start every recording session with an on-tape assignment of the training rights to the session:

https://pluralistic.net/2023/02/09/ai-monkeys-paw/#bullied-schoolkids

And now, with total predictability, Universal – the largest music company in the world – has announced that it will start training voice-clones with the music in its catalog:

https://www.rollingstone.com/music/music-news/umg-startsai-voice-clone-partnership-with-soundlabs-1235041808/

This comes hot on the heels of a massive blow-up between Universal and Tiktok, in which Universal professed its outrage that Tiktok was going to train voice-clones with the music Universal licensed to it. In other words: Universal's copyright claims over AI training cash out to this: "If anyone is going to profit from immiserating musicians, it's going to be us, not Tiktok."

I understand why Universal would like this idea. I just don't understand why any musician would root for Universal to defeat Tiktok, or Getty Images to trounce Stable Diffusion. Do you really think that Getty Images likes paying photographers and wants to give them a single penny more than they absolutely have to?

As we learned from George Orwell's avant-garde animated agricultural documentary Animal Farm, the problem isn't who holds the whip, the problem is the whip itself:

The creatures outside looked from pig to man, and from man to pig, and from pig to man again; but already it was impossible to say which was which.

Entertainment execs and tech execs alike are obsessed with AI because they view the future of "content" as fundamentally passive. Here's Ryan Broderick putting it better than I ever could:

At a certain audience size, you just assume those people are locked in and will consume anything you throw at them. Then it just becomes a game of lowering your production costs and increasing your prices to increase your margins. This is why executives love AI and why the average American can’t afford to eat at McDonald’s anymore.

https://www.garbageday.email/p/ceo-passive-content-obsession

Here's a rule of thumb for tech policy prescriptions. Any time you find yourself, as a worker, rooting for the same policy as your boss, you should check and make sure you're on the right side of history. The fact that creative bosses are so obsessed with making copyright cover more kinds of works, restrict more activities, lasting longer and generating higher damages should make creative workers look askance at these proposals.

After 40 years of expanded copyright, we have a creative industry that's larger and more profitable than ever, and yet the share of income going to creative workers has been in steady decline over that entire period. Every year, the share of creative income that creative workers can lay claim to declines, both proportionally and in real terms.

As with the mystery of Spotify's payments, this isn't a mystery at all. You just need to understand that when creators are stuck bargaining with a tiny, powerful cartel of movie, TV, music, publishing, streaming, games or app companies, it doesn't matter how much copyright they have to bargain with. Giving a creative worker more copyright is like giving a bullied schoolkid more lunch-money. There's no amount of money that will satisfy the bullies and leave enough left over for the kid to buy lunch. They just take everything.

Telling creative workers that they can solve their declining wages with more copyright is a denial that creative workers are workers at all. It treats us as entrepreneurial small businesses, LLCs with MFAs negotiating B2B with other companies. That's how we lose.

On the other hand, if we address the problems of AI and labor as workers, and insist on labor rights – like the Writers Guild did when it struck last summer – then we ally ourselves with every other worker whose wages and working conditions are being attacked with AI:

https://pluralistic.net/2023/10/01/how-the-writers-guild-sunk-ais-ship/

Our path to better working conditions lies through organizing and striking, not through helping our bosses sue other giant mulitnational corporations for the right to bleed us out.

The US Copyright Office has repeatedly stated that AI-generated works don't qualify for copyrights, meaning everything AI generated can be freely copied and distributed and the companies that make them can't stop them. This is fantastic news, because the only thing our bosses hate more than paying us is not being able to stop other people from copying the things we make for them. We should be shouting this from the rooftops, not demanding more copyright for AI.

Here's a thing: FTC chair Lina Khan recently told an audience that she was thinking of using her Section 5 powers (to regulate "unfair and deceptive" conduct) to go after AI training:

https://www.youtube.com/watch?v=3mh8Z5pcJpg

Khan has already used these Section 5 powers to secure labor rights, for example, by banning noncompetes:

https://pluralistic.net/2024/04/25/capri-v-tapestry/#aiming-at-dollars-not-men

Creative workers should be banding together with other labor advocates to propose ways for the FTC to prevent all AI-based labor exploitation, like the "reverse-centaur" arrangement in which a human serves as an AI's body, working at breakneck pace until they are psychologically and physically ruined:

https://pluralistic.net/2022/04/17/revenge-of-the-chickenized-reverse-centaurs/

As workers standing with other workers, we can demand the things that help us, even (especially) when that means less for our bosses. On the other hand, if we confine ourselves to backing our bosses' plays, we only stand to gain whatever crumbs they choose to drop at their feet for us.

If you'd like an essay-formatted version of this post to read or share, here's a link to it on pluralistic.net, my surveillance-free, ad-free, tracker-free blog:

https://pluralistic.net/2024/06/21/off-the-menu/#universally-loathed

Support me this summer on the Clarion Write-A-Thon and help raise money for the Clarion Science Fiction and Fantasy Writers' Workshop!

Image: Cryteria (modified) https://commons.wikimedia.org/wiki/File:HAL9000.svg

CC BY 3.0 https://creativecommons.org/licenses/by/3.0/deed.en