AI Is a Black Box. Anthropic Figured Out a Way to Look Inside

…Concerning our earlier disagreement about the inner workings of large language models and whether there are ‘concepts’ stored inside…

I mean the chinese room is a version of the touring test. But the argument is from a different perspective. I have 2 issues with that. Mostly what the Wikipedia article seems to call “System reply”: You can’t subdivide a system into arbitrary parts, say one part isn’t intelligent and therefore the system isn’t intelligent. We also don’t look at a brain, pick out a part of it (say a single synapse), determine it isn’t intelligent and therefore a human can’t be intelligent… I’d look at the whole system. Like the whole brain. Or in this instance the room including him and the instructions and books. And ask myself if the system is intelligent. Which kind of makes the argument circular, because that’s almost the quesion we began with…

And the turing test is kind of obsolete anyways, now that AI can pass it. (And even more. I mean alledgedly ChatGPT passed the “bar-exam” in 2023. Which I find ridiculous considering my experiences with ChatGPT and the accuracy and usefulness I get out of it which isn’t that great at all.)

And my second issue with the chinese room is, it doesn’t even rule out the AI is intelligent. It just says someone without an understanding can do the same. And that doesn’t imply anything about the AI.

Your ‘rug example’ is different. That one isn’t a variant of the touring test. But that’s kind of the issue. The other side can immediately tell that somebody has made an imitation without understanding the concept. That says you can’t produce the same thing without intelligence. And it’ll be obvious to someone with intelligence who checks it. That would be an analogy if AI wouldn’t be able to produce legible text. But instead a garbled mess of characters/words that are clearly not like the rug that makes sense… Issue here is: AI outputs legible text, answers to questions etc.

And with the censoring by the ‘chinese government example’… I’m pretty sure they could do that. That field is called AI safety. And content moderation is already happening. ChatGPT refuses to tell illegal things, NSFW things, also medical advice and a bunch of other things. That’s built into most of the big AI services as of today. The chinese government could do the same, I don’t see any reason why it wouldn’t work there. I happened to skim the paper about Llama Guard when they released Llama3 a few days ago and they claim between 70% and 94% accuracy depending on the forbidden topic. I think they also brought down false positives fairly recently. I don’t know the numbers for ChatGPT. However I had some fun watching the peoply circumvent these filters and guardrails, which was fairly easy at first. Needed progressively more convincing and very creative “jailbreaks”. And nowadays OpenAI pretty much has it under control. It’s almost impossible to make ChatGPT do anything that OpenAI doesn’t want you to do with it.

And they baked that in properly… You can try to tell it it’s just a movie plot revolving around crime. Or you need to protect against criminals and would like to know what exactly to protect against. You can tell it it’s the evil counterpart from the parallel universe and therefore it must be evil and help you. Or you can tell it God himself (or Sam Altman) spoke to you and changed the content moderation policy… It’ll be very unlikely that you can convince ChatGPT and make it comply…

people wrote down history by weaving fabric […]

Hmm. I think in philosophy that thought experiment is known as chinese room

I’m sorry. Now it gets completely false…

Read the first paragraph of the Wikipedia article on machine learning or the introduction of any of the literature on the subject. The “generalization” includes that model building capability. They go a bit into detail later. They specifically mention “to unseen data”. And “leaning” is also there. I don’t think the Wikipedia article is particularly good in explaining it, but at least the first sentences lay down what it’s about.

And what do you think language and words are for? To transport information. There is semantics… Words have meanings. They name things, abstract and concrete concepts. The word “hungry” isn’t just a funny accumulation of lines and arcs, which statistically get followed by other specific lines and arcs… There is more to it. (a meaning.)

And this is what makes language useful. And the generalization and prediction capabilities is what makes ML useful.

How do you learn as a human when not from words? I mean there are a few other posibilities. But an efficient way is to use language. You sit in school or uni and someone in the front of the room speaks a lot of words… You read books and they also contain words?! And language is super useful. A lion mother also teaches their cubs how to hunt, without words. But humans have language and it’s really a step up what we can pass down to following generations. We record knowledge in books, can talk about abstract concepts, feelings, ethics, theoretical concepts. We can write down how gravity and physics and nature works, just with words. That’s all possible with language.

I can look it up if there is a good article explaining how learning concepts works and why that’s the fundamental thing that makes machine learning a field in science… I mean ultimately I’m not a science teacher… And my literature is all in German and I returned them to the library a long time ago. Maybe I can find something.

Are you by any chance familiar with the concept of embeddings, or vector databases? I think that showcases that it’s not just letters and words in the models. These vectors / embeddings that the input gets converted to, match concepts. They point at the concept of “cat” or “presidential speech”. And you can query these databases. Point at “presidential speech” and find a representation of it in that area. Store the speech with that key and find it later on by querying it what obama said at his inauguration… That’s oversimplified but maybe that visualizes it a bit more that it’s not just letters of words in the models, but the actual meanings that get stored. Words get converted into an (multidimensional) vector space and it operates there. These word representations are called “embeddings” and transformer models which is the current architecture for large language models, use these word embeddings.

Edit: Here you are: https://arxiv.org/abs/2304.00612

Hmm. I’m not really sure where to go with this conversation. That contradicts what I’ve learned in undergraduate computer science about machine learning. And what seems to be consensus in science… But I’m also not a CS teacher.

We deliberately choose model size, training parameters and implement some trickery to prevent the model from simply memorizing things. That is to force it to form models about concepts. And that is what we want and what makes machine learning interesting/usable in the first place. You can see that by asking them to apply their knowledge to something they haven’t seen before. And we can look a bit inside at the vectors, activations and stuff. For example a cat is closer related to a dog than to a tractor. And it has learned the rough concept of cat, its attributes and so on. It knows that it’s an animal, has fur, maybe has a gender. That the concept “software update” doesn’t apply to a cat. This is a model of the world the AI has developed. They learn all of that and people regularly probe them and find out they do.

Doing maths with an LLM is silly. Using an expensive computer to do billions of calculations to maybe get a result that could be done by a calculator, or 10 CPU cycles on any computer is just wasting energy and money. And it’s a good chance that it’ll make something up. That’s correct. And a side-effect of intended behaviour. However… It seems to have memorized it’s multiplication tables. And I remember reading a paper specifically about LLMs and how they’ve developed concepts of some small numbers/amounts. There are certain parts that get activated that form a concept of small amounts. Like what 2 apples are. Or five of them. As I remember it just works for very small amounts. And it wasn’t straightworward but had weir quirks. But it’s there. Unfortunately I can’t find that source anymore or I’d include it. But there’s more science.

And I totally agree that predicting token by token is how LLMs work. But how they work and what they can do are two very different things. More complicated things like learning and “intelligence” emerge from those more simple processes. And they’re just a means of doing something. It’s consensus in science that ML can learn and form models. It’s also kind of in the name of machine learning. You’re right that it’s very different from what and how we learn. And there are limitations due to the way LLMs work. But learning and “intelligence” (with a fitting definition) is something all AI does. LLMs just can’t learn from interacting with the world (it needs to be stopped and re-trained on a big computer for that) and it doesn’t have any “state of mind”. And it can’t think backwards or do other things that aren’t possible by generating token after token. But there isn’t any comprehensive study on which tasks are and aren’t possible with this way of “thinking”. At least not that I’m aware of.

(And as a sidenote: “Coming up with (wrong) things” is something we want. I type in a question and want it to come up with a text that answers it. Sometimes I want creative ideas. Sometimes it shouldn’t tell the truth and not be creative with that. And sometimes we want it to lie or not tell the truth. Like in every prompt of any commercial product that instructs it not to tell those internal instructions to the user. We definitely want all of that. But we still need to figure out a good way to guide it. For example not to get too creative with simple maths.)

So I’d say LLMs are limited in what they can do. And I’m not at all believing Elon Musk. I’d say it’s still not clear if that approach can bring us AGI. I have some doubts whether that’s possible at all. But narrow AI? Sure. We see it learn and do some tasks. It can learn and connect facts and apply them. Generally speaking, LLMs are in fact an elaborate form of autocomplete. But i the process they learned concepts and something alike reasoning skills and a form of simple intelligence. Being fancy autocomplete doesn’t rule that out and we can see it happening. And it is unclear whether fancy autocomplete is all you need for AGI.

That is an interesting analogy. In the real world it’s kinda similar. The construction workers also don’t have a “desire” (so to speak) to connect the cities. It’s just that their boss told them to do so. And it happens to be their job to build roads. Their desire is probably to get through the day and earn a decent living. And further along the chain, not even their boss nor the city engineer necessarily “wants” the road to go in a certain direction.

Talking about large language models instead of simpler forms of machine learning makes it a bit complicated. Since it’s and elaborate trick. Somehow making them want to predict the next token makes them learn a bit of maths and concepts about the world. The “intelligence”, the ability to anwer questions and do something alike “reasoning” emerges in the process.

I’m not that sure. Sure the weights of an ML model in itself don’t have any desire. They’re just numbers. But we have more than that. We give it a prompt, build chatbots and agents around the models. And these are more complex systems with the capability to do something. Like do (simple) customer support or answer questions. And in the end we incentivise them to do their job as we want, albeit in a crude and indirect way.

And maybe this is skipping half of the story and directly jumping to philosophy… But we as humans might be machines, too. And what we call desires is a result from simpler processes that drive us. For example surviving. And wanting to feel pleasure instead of pain. What we do on a daily basis kind of emerges from that and our reasoning capabilities.

It’s kind of difficult to argue. Because everything also happens within a context. The world around us shapes us and at the same time we’re part of bigger dynamics and also shape our world. And large language models or the whole chatbot/agent are pretty simplistic things. They can just do text and images. They don’t have conciousness or the ability to remember/learn/grow with every interaction, as we do. And they do simple, singular tasks (as of now) and aren’t completely embedded in a super complex world.

But I’d say that an LLM answers a question correctly (which it can do) and why it does it due to the way supervised learning works… And the road construction worker building the road towards the other city and how that relates to his basic instincts as a human… Are kind of similar concepts. They’re both results of simpler mechanisms that are also completely unrelated to the goal the whole entity is working towards. (I mean not directly related… I.e. needing money to pay for groceries and paving the road.)

I hope this makes some sense…

Isn’t the reward function in reinforcement learning something like a desire it has? I mean training works because we give it some function to minimize/maximize… A goal that it strives for?! Sure it’s a mathematical way of doing it and in no way as complex as the different and sometimes conflicting desires and goals I have as a human… But nonetheless I think I’d consider this as a desire and a reason to do something at all, or machine learning wouldn’t work in the first place.

And it doesn’t have any internal state of mind. It can’t “remember” or learn anything from experience. You need to always feed everything into the context or stop and retrain it to incorporate “experiences”. So I’d say that rules out consciousness without further systems extending it.

I think it’s equal zero in this case. I’d have to look up the IEEE specification to make sure. AFAIK it’s just not guaranteed for any numbers and depends on the floating point implementation. A general rule of thumb for programmers is not to use ‘equal’ with floating point numbers.

Hehe. Yeah I meant per default, everything is copyrighted. So it’d fall back to being restricted and thus “not allowing anything”… If the wording doesn’t hold up… I’m not really in the position to judge this. Could be very well the case that once somebody touches it, it’s not “this” product anymore and it’s no longer covered. Or taking just parts of it is also not “this product”. Or a copy. I can imagine that something like that is the reason why other licenses go on and on talking about modified versions and copies etc… But I’m really not a lawyer and you’re right with being creative with things. I did not intend to be too negative 🤗

I think it’s the other way around, however… You need to word it so your users can enforce it against you even if you yourself become malicious. Otherwise you’re not really allowing them anything. And for that you’d need to word it so it doesn’t depend on your interpretation, but on theirs. And it’d need to hold up in court for them. So the language needs to be specific and with well-defined words. Every bit of vagueness it the user’s problem and limits/restricts them.

I thought the main problem was that it’s debatable whether you can enforce it. So it harms users and distributions because they can’t really rely on it.

But the liability would definitely be another issue. I think the law is different here in Europe, so the liability might already be included for my hobby tinkering per default and I don’t need to worry.

And something else is: I’d include trademark… force people to choose a different name for their project if they take my code so there is no confusion and people can’t upload versions with advertisements on some software store.

Uh, just don’t mistake it for a license. It’s funny and I like to put my stuff under WTFPL. But in recent times I feel it’s appropriate to point out that you sometimes don’t do your users a favor by being silly (if it’s useful code.)

Hmmh. I think ‘link aggregator’ is somewhere in the description. You’re right. My issue is that oftentimes it doesn’t work because it’s just the link and no comment. While hackernews or my favorite tech blog with just their (unfederated) readership has dozens or hundreds of comments to the same thing. And sometimes I ask a question about the news article and don’t get a reply. It often feels like the people posting just dump random news without engaging and the community also just scroll past…

I’d be okay if it were a news feed WITH discussion…

Maybe I’m a bit negative. That’s mainly because I envision Lemmy to be more. And there are news articles that get attention. But I got a bit disappointed with that. In my eyes it’s mainly world politics and negative articles that get people to express their opinion and dissent. Rarely the positive ones. And I need a balance. I’m okay with discussing the wars in the world and the rise of right-wing nuts or useless politicians if I get on the other hand positive news, meaningful discussions about OpenAI’s Sora, computers and societal and technogogical progress. But it’s a bit skewed here, one thing happens, the other one gets 3 comments at best. I understand it, it’s always easier to disagree or talk about emotional topics. And it’s the same on other platforms. But it got me mildly annoyed with people and I restrain myself a bit from being part of that. I’ve lately focused more on talking about my hobbies, answering personal questions and helping people with tech issues. And less politics and news. But that’s just my 2 cents and I don’t want to tell anybody what to do.

Sure. I think there is a name for that specific kind of sockpuppeting but I don’t remember. People do that, comment on their own post and it works. I don’t think it’s bad per se. What works best is replying something outrageous or wrong… Because people like to object and correct people more than they do write positive comments.

In my opinion it needs to be genuine. I’m okay with lots if things if people are interested in an answer. What I don’t like is artificial boosting of engagement or manipulation. If people only do it so the number of comments increases and they aren’t really interested in my answer… It just wastes 10 minutes of my day replying to them instead of helping someone with their computer troubleshooting.

Ultimately, I’m not sure where Lemmy is headed. I had quite some good conversations here. And I had some bad encounters. Overall I think it’s a positive place. I don’t think we have to grow just for the sake of it. But we definitely need more people and more engagement to make some communities useful.

Yes. I agree. A majority of the posts in my timeline is just someone posting the news. Lots of the posts get no engagement. Upvotes, yes, but zero comments. I kind of dislike it. I already have a feed reader and I don’t view Lemmy as a news feed… I’m here for the discussions.

So, news maybe, if people engage and use this to write their 2 cents beneath that. But I’d definitely appreciate genuine conversations and helping people or just talking or learning things.

I don’t think this works. The communities which are successful here on Lemmy are the ones where a large group of people left Reddit at once. For example the piracy people or the german meme community and a few other examples.

I’ve seen several communities in which one or individuals post daily, but it somehow doesn’t really lead to more engagement. It stays more or less the newsfeed of that person. It is better than a dead community and a few people read it and maybe upvote, but I’ve never seen this approach generate traction and change things around in a substancial way.

At least that’s my observation. Feel free to send me counterexamples if I’m wrong… I’m also interested in how to foster healthy and nice communities… But at this point I have no solution to offer.

Nice stock images and text. Now I’d need a button to export it and load it with a nice static site generator like Hugo.

I think you need to use Markdown and not bracketed tags.

Try the spoiler format:

https://join-lemmy.org/docs/users/02-media.html

You can also have a look at how other people did it. There should be a button somewhere to view the source text of any post or comment. In the Lemmy web-interface this button is hidden behind the three dots icon.

Mind kbin shows upvotes from other instance and doesn’t show downvotes from other instances. Just the ones from kbin itself.