First of all, I understand your point of view. And I’ve been looking at artists being undervalued like your potential client for decades, before AI was even a thing. So I definitely feel you on that point, and I wish it would be different. That said, here’s my response. (It’s a bit long, so I put it in spoiler tags)

I told him he wasn’t looking for a composer, but rather a programmer or something

_

I could also tell you about the written assignments that students hand in, and for which I can identify in less than 30 seconds which ones have been produced by AI (students overreact to their writing skills, it’s often laughable).

_

As I tell them, those who have used chatgpt have “learned” to use AI, those who have done the work have learned to carry out research, to synthesize their ideas and to structure, articulate and present them.

_

One last thing. As far as innovation is concerned, AI can endlessly produce pieces that sound like Bach, but it took Bach to exist in the first place, and Glenn Gould to revolutionize the interpretation of his scores for this to be possible.

_

EDIT: Just want to add to this:

I have no interest in replacing this practice by entering prompts into an algorithm, even if I could make easy money from it.

That’s not something anyone should do. Because that’s not using it as a tool. That’s making it the entire process. That’s not the kind of AI usage I’m advocating for either. And you’re free to forego AI completely. Just like there are probably some instruments you never use, or some genre you never visit. I don’t like taking the easy way either, that’s why I make creative stuff as a living too. If I just wanted money I would go elsewhere too.

Again - if this is your argument - then the vast majority of things humans do would be “pushing down harder on the gas pedal”. Excluding AI, more people get born every year, water usage also increases every year, electricity usage too. Even if you got rid of all AI right now you would have to overcome those much more significant increases to make a difference. It just doesn’t even make a dent. And it has to, if you want it to actually reduce the impact of climate change and resource depletion.

The world does not stand still, even if we did everything we should to stop climate change. Technology that can change the world and facilitates happier, healthier humans is not a bad thing for a reasonable price. And as I just explained in detail, that price is not that significant in the grand scheme of things. Hence why there is no significant public outrage from this.

If you’re going to hold this position, you should really stick to the biggest polluters, which as you agree, are not getting enough pushback. I agree with that as well, and I would happily stand by your side here. But if someone is handing out pie, and you think everyone should be angry at someone taking 0.00532% of the pie, that is horribly ineffective at actually getting the change we need. Since basically nobody reasonable is going to agree with you. While for the larger polluters it is easily self evident they need it, and we still have a lot of trouble with that.

And Sam Altman went to the emirates asking for Trillions to scale LLMs. All of this for a little more convenience when tackling mostly mundane tasks.

I don’t know how many times I have to say this, but I don’t like the big tech companies use of AI. That does not say anything about the technology at large though. Screw OpenAI and Sam Altman. If your criticism is purely aimed at wasteful conduct by big companies, I’m all there with you. But there are so many smaller companies that also use AI and LLMs.

If you think that’s depressing, wait until you find out that it’s basically nothing in the grand scheme of things.

_

And as others have pointed out the water isn’t gone, there’s some cyclicality in how the water is used.

I already know of the stats you speak of, and it’s not nothing. But as I explain below, on the scale of other pollutants and for how widespread it’s usage is, it is a footnote. There are far bigger fish to fry first when it comes to reducing water usage and energy usage.

Lets look up some stats together. Most sources agree that we use about 4 trillion cubic meters of water every year worldwide (Although, this stat is from 2015 most likely, and so it will be bigger now). In 2022, using the stats here Microsoft used 1.7 billion gallons per year, and Google 5.56 billion gallons per year. In cubic meters that’s only 23.69 million cubic meters. That’s only 0.00059% of the worldwide water usage. Meanwhile agriculture uses 70%. Granted not every country uses 70%, but a 1% gain there overshadows any current and even future usage.

And even if we just look at the US, since that’s where Google and Microsoft are based, which uses 322 billion gallons of water every day, resulting in about 445 billion cubic meters per year, that’s still 0.00532%. So it’s not hugely significant inside the US either, we can have 187 more Googles and Microsofts before we even top a single percentage.

There are plenty of other hobbies that are also terrible for the environment but we don’t tackle them because like AI it would be a drop in the bucket. Honestly ask yourself, if AI was such a big issue and would threaten lives, why is nobody actually taking serious measures for it? Why are there no public campaigns to get people to stop using AI to save the water? It’s because the stats are concerning, but not significantly so, and if the stats don’t actually back up the position that AI usage is slowly sucking us dry, people don’t stand behind it. If the advancement of AI allows us to save more lives through medical research, or allows us to farm more efficiently, it can easily pay back it’s investment. And those things outweigh the negatives.

However, as I’ve stated multiple times below, I despise the wasteful usage of AI, and therefore you won’t see me praising Bing or Google for their usage of AI in a way that makes no sense and just wastes resources. But the technology exists for everyone, not just those companies. There are more sustainable LLM models than GPT-4, and OpenAI can rightfully be criticized for not prioritizing efficiency rather that bigger and more power hungry models.

EDIT: Added US comparison EDIT2: Double checked some of the math

Do feel free to explain what you mean and why, because this comment literally has nothing for me to even respond to.

EDIT: And if you meant the water cost, you can read my rationale on that here.

Most things produced by AI and assisted by AI are still human creation, as it requires a human to guide it to what it’s making. Human innovation is also very much based on mixing materials it’s seen before in new creative manners. Almost no material is truly innovative. Ask any honest artists about their inspirations and they can tell you what parts of their creations were inspired by what. Our world has explored the depths of most art forms so there is more than a lifetime’s worth of art to mix and match. Often the real reason things feel fresh and new is because they are fresh and new to us, but already existed in some form out there before it came to our attention.

That AI can match this is easily proven by fact AI can create material that no human would realistically make (like AI generated QR codes, or ‘cursed’ AI), very proficient style mixing that would take a human extensive study of both styles to pull off (eg. Pokemon and real life), or real looking images that could not realistically, financially, conscionably, be made using normal methods (eg. A bus full of greek marble statues).

Nobody is saying you have to like AI art, and depending on your perspective, some or most of it will still be really low effort and not worth paying attention to, but that was already the state of art before AI. Lifetimes of art are being uploaded every day, but nobody has the time to view it all. So I would really keep an open mind that good AI art and AI assisted art exists out there, and you might one day come to like it but not realize you’re seeing it, because good AI usage is indistinguishable from normal art.

And Google as one of the largest companies in the world should be held fully responsible for that. But Google isn’t everyone. Google also has a huge amount of computers under their name, but does that mean everyone with a computer should be held to the same standards because it’s the same technology?

If sweeping conclusions are going to be made about the technology, it has to be looked at outside of the context of a specific company and how they implement and use AI. Otherwise, it should be specified that this is a criticism of specific companies and how they use AI, and I’d be totally there agreeing with you in the case of Google.

The earth has been warming for decades at this point, even before AI. We know what causes climate change and AI has so far been a footnote on a large list of unsolved problems. The moment that changes I’ll be right there with you, but I’m far more interested in taking down the companies that are largely responsible for it. This logic follows that just living life as a human is unethical because you can’t be environmentally neutral in today’s society, and I reject that notion.

Yes, exactly. And I don’t even disagree with making things better on the environment. It’s why I dislike LLMs being pushed into random things that don’t really need it. And if more efficient models exist, they should preferably be used.

But using AI to make your life a little better also brings positives that outweigh that cost. And it seems like as a society we have much bigger polluters to take care of that use orders more electricity and water, and we already struggle massively doing just that. So it really feels like misdirected anger towards the unfairness in society (which the societal cost they mentioned also seems to point to) than a real criticism of AI. And I can understand that anger, but not the mindset behind the conclusions derived from it.

This kind of AI approaches art in a way that finally kinda makes sense for my brain, so it’s frustrating seeing it shot down by people who don’t actually understand it. Stop using this stuff for tasks it wasn’t meant for (unless it’s a novelty “because we could” kind of way) and it becomes a lot more palatable.

Preach! I’m surprised to hear it works for people with aphantasia too, and that’s awesome. I personally have a very vivid mind’s eye and I can often already imagine what I want something to look like, but could never put it to paper in a satisfying way that didn’t cost excruciating amount of time. GenAI allows me to do that with still a decent amount of touch up work, but in a much more reasonable timeframe. I’m making more creative work than I’ve ever been because of it.

It’s crazy to me that some people at times completely refuse to even acknowledge such positives about the technology, refuse to interact with it in a way that would reveal those positives, refuse to look at more nuanced opinions of people that did interact with it, refuse even simple facts about how we learn and interact with other art and material, refusing legal realities like the freedom to analyze that allow this technology to exist (sometimes even actively fighting to restrict those legal freedoms, which would hurt more artists and creatives than it would help, and give even more more power to corporations and those with enough capital to self sustain AI model creation).

It’s tiring, but luckily it seems to be mostly an issue on the internet. Talking to people (including artists) in real life about it shows that it’s a very tiny fraction that holds that opinion. Keep creating 👍

Totally second the latter part - it’s the self destructive nature of being blindly anti-AI. Pretty much everyone would support giving more rights and benefits to people displaced by AI, but only a fraction of that group would support an anti-AI mentality. If you want to work against the negative effects of AI in a way that can actually change things, the solution is not to push against the wall closing in on you, but to find the escape.

This is like saying you can’t play video games because it costs electricity and you can go without. You can say it about literally everything that isn’t strictly necessary to live. AI isn’t just LLMs and only LLMs have a high environmental cost, and unless you are literally wasting the output like the big tech companies are, even that can be justified for the right reasons.

Yeah and honestly, this is largely a reasonable standard for anyone running an email server. If you don’t have SPF, DKIM and DMARC, basically anyone can spoof your emails and you’d be none the wiser. It also makes spam much harder to send without well, sacrificing IP addresses to the many spam lists. I wouldn’t be surprised if some people setting up their own mail server were made aware of these things because of being blocked.

There is so much wrong with this…

AI is a range of technologies. So yes, you can make surveillance with it, just like you can with a computer program like a virus. But obviously not all computer programs are viruses nor exist for surveillance. What a weird generalization. AI is used extensively in medical research, so your life might literally be saved by it one day.

You’re most likely talking about “Chat Control”, which is a controversial EU proposal to scan either on people’s devices or from provider’s ends for dangerous and illegal content like CSAM. This is obviously a dystopian way to achieve that as it sacrifices literally everyone’s privacy to do it, and there is plenty to be said about that without randomly dragging AI into that. You can do this scanning without AI as well, and it doesn’t change anything about how dystopian it would be.

You should be using end to end regardless, and a VPN is a good investment for making your traffic harder to discern, but if Chat Control is passed to operate on the device level you are kind of boned without circumventing this software, which would potentially be outlawed or made very difficult. It’s clear on it’s own that Chat Control is a bad thing, you don’t need some kind of conspiracy theory about ‘the true purpose of AI’ to see that.

I never anthropomorphized the technology, unfortunately due to how language works it’s easy to misinterpret it as such. I was indeed trying to explain overfitting. You are forgetting the fact that current AI technology (artificial neural networks) are based on biological neural networks. There is a range of quirks that it exhibits that biological neural networks do as well. But it is not human, nor anything close. But that does not mean that there are no similarities that can be rightfully pointed out.

Overfitting isn’t just what you describe though. It also occurs if the prompt guides the AI towards a very specific part of it’s training data. To the point where the calculations it will perform are extremely certain about what words come next. Overfitting here isn’t caused by an abundance of data, but rather a lack of it. The training data isn’t being produced from within the model, but as a statistical inevitability of the mathematical version of your prompt. Which is why it’s tricking the AI, because an AI doesn’t understand copyright - it just performs the calculations. But you do. And so using that as an example is like saying “Ha, stupid gun. I pulled the trigger and you shot this man in front of me, don’t you know murder is illegal buddy?”

Nobody should be expecting a machine to use itself ethically. Ethics is a human thing.

People that use AI have an ethical obligation to avoid overfitting. People that produce AI also have an ethical obligation to reduce overfitting. But a prompt quite literally has infinite combinations (within the token limits) to consider, so overfitting will happen in fringe situations. That’s not because that data is actually present in the model, but because the combination of the prompt with the model pushes the calculation towards a very specific prediction which can heavily resemble or be verbatim the original text. (Note: I do really dislike companies that try to hide the existence of overfitting to users though, and you can rightfully criticize them for claiming it doesn’t exist)

This isn’t akin to anything human, people can’t repeat pages of text verbatim like this and no toddler can be tricked into repeating a random page from a random book as you say.

This is incorrect. A toddler can and will verbatim repeat nursery rhymes that it hears. It’s literally one of their defining features, to the dismay of parents and grandparents around the world. I can also whistle pretty much my entire music collection exactly as it was produced because I’ve listened to each song hundreds if not thousands of times. And I’m quite certain you too have a situation like that. An AI’s mind does not decay or degrade (Nor does it change for the better like humans) and the data encoded in it is far greater, so it will present more of these situations in it’s fringes.

but it isn’t crafting its own sentences, it’s using everyone else’s.

How do you think toddlers learn to make their first own sentences? It’s why parents spend so much time saying “Papa” or “Mama” to their toddler. Exactly because they want them to copy them verbatim. Eventually the corpus of their knowledge grows big enough to the point where they start to experiment and eventually develop their own style of talking. But it’s still heavily based on the information they take it. It’s why we have dialects and languages. Take a look at what happens when children don’t learn from others: https://en.wikipedia.org/wiki/Feral_child So yes, the AI is using it’s training data, nobody’s arguing it doesn’t. But it’s trivial to see how it’s crafting it’s own sentences from that data for the vast majority of situations. It’s also why you can ask it to talk like a pirate, and then it will suddenly know how to mix in the essence of talking like a pirate into it’s responses. Or how it can remember names and mix those into sentences.

Therefore it is factually wrong to state that it doesn’t keep the training data in a usable format

If your arguments is that it can produce something that happens to align with it’s training data with the right prompt, well yeah that’s not incorrect. But it is so heavily misguided and borders bad faith to suggest that this tiny minority of cases where overfitting occurs is indicative of the rest of it. LLMs are a prediction machines, so if you know how to guide it towards what you want it to predict, and that is in the training data, it’s going to predict that most likely. Under normal circumstances where the prompt you give it is neutral and unique, you will basically never encounter overfitting. You really have to try for most AI models.

But then again, you might be arguing this based on a specific AI model that is very prone to overfitting, while I am arguing this out of the technology as a whole.

This isn’t originality, creativity or anything that it is marketed as. It is storing, encoding and copying information to reproduce in a slightly different format.

It is originality, as these AI can easily produce material never seen before in the vast, vast majority of situations. Which is also what we often refer to as creativity, because it has to be able to mix information and still retain legibility. Humans also constantly reuse phrases, ideas, visions, ideals of other people. It is intellectually dishonest to not look at these similarities in human psychology and then treat AI as having to be perfect all the time, never once saying the same thing as someone else. To convey certain information, there are only finite ways to do so within the English language.

This is an issue for the AI user though. And I do agree that needs to be more conscious in people’s minds. But I think time will change that. Perhaps when the photo camera came out there were some shmucks that took pictures of people’s artworks and claimed it as their own because the novelty of the technology allowed that for a bit, but eventually those people are properly differentiated from people properly using it.

Like if I download a textbook to read for a class instead of buying it - I could be proscecuted for stealing

Ehh, no almost certainly not (But it does depend on your local laws). But that honestly just sounds like some corporate boogyman to prevent you from pirating their books. The person hosting the download, if they did not have the rights to publicize it freely, would possibly be prosecuted though.

To illustrate, there’s this story of John Cena who sold a special Ford after signing a contract with Ford to explicitly forbid him from doing that. However, the person who bought the car was never prosecuted or sued, because they received the car from Cena with no strings attached. They couldn’t be held responsible for Cena’s break of contract, but Cena was held personally responsible by Ford.

For physical goods there is ‘theft by proxy’ though (receiving stolen goods that you know are most likely stolen), but that quite certainly doesn’t apply to digital, copyable goods. As to even access any kind of information on the internet, you have to download and thus, copy it.

That would be true if they used material that was paywalled. But the vast majority of the training information used is publicly available. There’s plenty of freely available books and information that you only require an internet connection for to access, and learn from.

Your first point is misguided and incorrect. If you’ve ever learned something by ‘cramming’, a.k.a. just repeating ingesting material until you remember it completely. You don’t need the book in front of you anymore to write the material down verbatim in a test. You still discarded your training material despite you knowing the exact contents. If this was all the AI could do it would indeed be an infringement machine. But you said it yourself, you need to trick the AI to do this. It’s not made to do this, but certain sentences are indeed almost certain to show up with the right conditioning. Which is indeed something anyone using an AI should be aware of, and avoid that kind of conditioning. (Which in practice often just means, don’t ask the AI to make something infringing)

This would be a good point, if this is what the explicit purpose of the AI was. Which it isn’t. It can quote certain information verbatim despite not containing that data verbatim, through the process of learning, for the same reason we can.

I can ask you to quote famous lines from books all day as well. That doesn’t mean that you knowing those lines means you infringed on copyright. Now, if you were to put those to paper and sell them, you might get a cease and desist or a lawsuit. Therein lies the difference. Your goal would be explicitly to infringe on the specific expression of those words. Any human that would explicitly try to get an AI to produce infringing material… would be infringing. And unknowing infringement… well there are countless court cases where both sides think they did nothing wrong.

You don’t even need AI for that, if you followed the Infinite Monkey Theorem and just happened to stumble upon a work falling under copyright, you still could not sell it even if it was produced by a purely random process.

Another great example is the Mona Lisa. Most people know what it looks like and if they had sufficient talent could mimic it 1:1. However, there are numerous adaptations of the Mona Lisa that are not infringing (by today’s standards), because they transform the work to the point where it’s no longer the original expression, but a re-expression of the same idea. Anything less than that is pretty much completely safe infringement wise.

You’re right though that OpenAI tries to cover their ass by implementing safeguards. Which is to be expected because it’s a legal argument in court that once they became aware of situations they have to take steps to limit harm. They can indeed not prevent it completely, but it’s the effort that counts. Practically none of that kind of moderation is 100% effective. Otherwise we’d live in a pretty good world.