Yea, try talking to chatgpt about things that you really know in detail about. It will fail to show you the hidden, niche things (unless you mention them yourself), it will make lots of stuff up that you would not pick up on otherwise (and once you point it out, the bloody thing will “I knew that” you, sometimes even if you are wrong) and it is very shallow in its details. Sometimes, it just repeats your question back to you as a well-written essay. And that’s fine…it is still a miracle that it is able to be as reliable and entertaining as some random bullshitter you talk to in a bar, it’s good for brainstorming too.
It’s like watching mainstream media news talk about something you know about.
Oh good comparison
Haha, definitely, it’s infuriating and scary. But it also depends on what you are watching for. If you are watching TV, you do it for convenience or entertainment. LLMs have the potential to be much more than that, but unless a very open and accessible ecosystem is created for them, they are going to be whatever our tech overlords decide they want them to be in their boardrooms to milk us.
Well, if you read the article, you’ll see that’s exactly what is happening. Every company you can imagine is investing the GDP of smaller nations into AI. Google, Facebook, Microsoft. AI isn’t the future of humanity. It’s the future of capitalist interests. It’s the future of profit chasing. It’s the future of human misery. Tech companies have trampled all over human happiness and sanity to make a buck. And with the way surveillance capitalism is moving—facial recognition being integrated into insane places, like the M&M vending machine, the huge market for our most personal, revealing data—these could literally be two horsemen of the apocalypse.
Advancements in tech haven’t helped us as humans in while. But they sure did streamline profit centers. We have to wrest control of our future back from corporate America because this plutocracy driven by these people is very, very fucking dangerous.
AI is not the future for us. It’s the future for them. Our jobs getting “streamlined” will not mean the end of work and the rise of UBI. It will mean stronger, more invasive corporations wielding more power than ever while more and more people suffer, are cast out and told they’re just not working hard enough.
What worries me is how much of the AI criticism on Lemmy wants to make everything worse; not share the gains more equally. If that’s what passes for left today, well…
I don’t think they have that much potential. They are just uncontrollable, it’s a neat trick but totally unreliable if there isn’t a human in the loop. This approach is missing all the control systems we have in our brains.
I really only use for “oh damn, I known there’s a great one-liner to do that in Python” sort of thing. It’s usually right and of it isn’t it’ll be immediacy obvious and you can move on with your day. For anything more complex the gas lighting and subtle errors make it unusable.
Oh yes, it’s great for that. My google-fu was never good enough to “find the name of this thing that does this, but only when in this circumstance”
ChatGPT is great for helping with specific problems. Google search for example gives fairly general answers, or may have information that doesn’t apply to your specific situation. But if you give ChatGPT a very specific description of the issue you’re running into it will generally give some very useful recommendations. And it’s an iterative process, you just need to treat it like a conversation.
It’s also a decent writer’s room brainstorm kind of tool, although it can’t really get beyond the initial pitch as it’s pretty terrible at staying consistent when trying to clean up ideas.
I find it incredibly helpful for breaking into new things.
I want to learn terraform today, no guide/video/docs site can do it as well as having a teacher available at any time for Q&A.
Aside from that, it’s pretty good for general Q&A on documented topics, and great when provided context (ie. A full 200MB export of documentation from a tool or system).
But the moment I try and dig deeper I to something I’m an expert in, it just breaks down.
That’s why I’ve found it somewhat dangerous to use to jump into new things. It doesn’t care about bes practices and will just help you enough to let you shoot yourself in the foot.
Just wait for MeanGirlsGPT
Good. It’s dangerous to view AI as magic. I’ve had to debate way too many people who think they LLMs are actually intelligent. It’s dangerous to overestimate their capabilities lest we use them for tasks they can’t perform safely. It’s very powerful but the fact that it’s totally non deterministic and unpredictable means we need to very carefully design systems that rely on LLMs with heavy guards rails.
Not being combative or even disagreeing with you - purely out of curiosity, what do you think are the necessary and sufficient conditions of intelligence?
A worldview simulation it can use as a scratch pad for reasoning. I view reasoning as a set of simulated actions to convert a worldview from state a to state b.
It depends on how you define intelligence though. Normally people define it as human like, and I think there are 3 primary sub types of intelligence needed for cognizance, being reasoning, awareness, and knowledge. I think the current Gen is figuring out the knowledge type, but it needs to be combined with the other two to be complete.
Thanks! I’m not clear on what you mean by a worldview simulation as a scratch pad for reasoning. What would be an example of that process at work?
For sure, defining intelligence is non trivial. What clear the bar of intelligence, and what doesn’t, is not obvious to me. So that’s why I’m engaging here, it sounds like you’ve put a lot of thought into an answer. But I’m not sure I understand your terms.
A worldview is your current representational model of the world around you, so for example you know you’re a human on earth in a physical universe when a set of rules, you have a mental representation of your body and it’s capabilities, your location and the physicality of the things in your location. It can also be abstract things too, like your personality and your relationships and your understanding of what’s capable in the world.
Basically, you live in reality, but you need a way to store a representation of that reality in your mind in order to be able to interact with and understand that reality.
The simulation part is your ability to imagine manipulating that reality to achieve a goal, and if you break that down, you’re trying to convert reality from your perceived current real state A, to a imagined desired state B. Reasoning is coming up with a plan to convert the worldview from state A to state B step by step, so let’s say you want to brush your teeth, you a want to convert your worldview of you having dirty teeth to you having clean teeth, and to do that you reason that you need to follow a few steps to achieve that, like moving your body to the bathroom, retrieving tools (toothbrush and toothpaste) and applying mechanical action to your teeth to clean them. You created a step by step plan to change the state of your worldview to a new desired state you came up with. It doesn’t need to be physical either, it could be an abstract goal, like calculating a tip for a bill. It can also be a grand goal, like going to college or creating a mathematical proof.
LLMs don’t have a representational model of the world, they don’t have a working memory or a world simulation to use as a scratchpad for testing out reasoning. They just take a sequence of words and retrieve the next word that is probabilistically and relationally likely to be a good next word based on its training data.
They could be a really important cortex that can assist in developing a worldview model, but in their current granular state of being a single task AI model, they cannot do reasoning on their own.
Knowledge retrieval is an important component that assists in reasoning though, so it can still play a very important role in reasoning.
Interesting. I’m curious to know more about what you think of training datasets. Seems like they could be described as a stored representation of reality that maybe checks the boxes you laid out. It’s a very different structure of representation than what we have as animals, but I’m not sure it can be brushed off as trivial. The way an AI interacts with a training dataset is mechanistic, but as you describe, human worldviews can be described in mechanistic terms as well (I do X because I believe Y).
You haven’t said it, so I might be wrong, but are you pointing to freewill and imagination as somehow tied to intelligence in some necessary way?
I think worldview is all about simulation and maintaining state, it’s not really about making associations, but rather maintaining some kind of up to date and imaginary state that you can simulate on top of, to represent the world. I think it needs to be a very dynamic thing which is a pretty different paradigm to the ML training methodology.
Yes, I view these things as foundational to freewill and imagination, but I’m trying to think more low level than that. Simulation facilities imagination and reasoning facilities motivation which facilities free will.
Are those things necessary for intelligence? Well it depends on your definition and everyone has a different definition ranging from reciting information to full blown consciousness. Personally, I don’t really care about coming up with a rigid definition for it, it’s just a word, I care more about the attributes. I think LLMs are a good knowledge engine and knowledge is a component of intelligence.
I think it’s a big mistake to think that because the most basic LLMs are just autocompletes, or that because LLMs can hallucinate, that what big LLMs do doesn’t constitute “thinking”. No, GPT4 isn’t conscious, but it very clearly “thinks”.
It’s started to feel to me like current AIs are reasonable recreations of parts of our minds. It’s like they’re our ability to visualize, to verbalize, and to an extent, to reason (at least the way we intuitively reason, not formally), but separared from the “rest” of our thought processes.
Depends on how you define thinking. I agree, LLMs could be a component of thinking, specifically knowledge and recall.
Yes, as Linus Torvalds said humans are also thinking like autocomplete systems.
There’s magic?
Only if you believe in it. Many CEOs do. They’re very good in magical thinking.
I have a counter argument. From an evolutionary standpoint, if you keep doubling computer capacity exponentially isn’t it extraordinarily arrogant of humans to assume that their evolutionarily stagnant brains will remain relevant for much longer?
If you keep doubling the number of fruit flies exponentially, isn’t it likely that humanity will find itself outsmarted?
The answer is no, it isn’t. Quantity does not quality make and all our current AI tech is about ways to breed fruit flies that fly left or right depending on what they see.
You can make the same argument about humans that you do AI, but from a biological and societal standpoint. Barring any jokes about certain political or geographical stereotypes, humans have gotten “smarter” that we used to be. We are very adaptable, and with improvements to diet and education, we have managed to stay ahead of the curve. We didn’t peak at hunter-gatherer. We didn’t stop at the Renaissance. And we blew right past the industrial revolution. I’m not going to channel my “Humanity, Fuck Yeah” inner wolf howl, but I have to give our biology props. The body is an amazing machine, and even though we can look at things like the current crop of AI and think, “Welp, that’s it, humans are done for,” I’m sure a lot of people thought the same at other pivotal moments in technological and societal advancement. Here I am, though, farting taco bell into my office chair and typing about it.
You can compare human intelligence to centuries ago on a simple linear scale. Neural density has not increased by any stretch of the imagination in the way that transistor density has. But I’m not just talking density I’m talking about scalability that is infinite. Infinite scale of knowledge and data.
Let’s face it people are already not that intelligent, we are smart enough to use the technology of other smarter people. And then there are computers, they are growing intelligently with an artificial evolutionary pressure being exerted on their development, and you’re telling me that that’s not going to continue to surpass us in every way? There is very little to stop computers from being intelligent on a galactic scale.
Computer power doesn’t scale infinitely, unless you mean building a world mind and powering if off of the spinning singularity at the center of the galaxy like a type 3 civilization, and that’s sci-fi stuff. We still have to worry about bandwidth, power, cooling, coding and everything else that going into running a computer. It doesn’t just “scale”. There is a lot that goes into it, and it does have a ceiling. Quantum computing may alleviate some of that, but I’ll hold my applause until we see some useful real world applications for it.
Furthermore, we still don’t understand how the mind works, yet. There are still secrets to unlock and ways to potentially augment and improve it. AI is great, and I fully support the advancement in technology, but don’t count out humans so quickly. We haven’t even gotten close to human level intelligence and GOFAI, and maybe we never will.
As I said that answer seems incredibly arrogant in the face of evolutionary pressure and logarithmic growth.
You can believe whatever you want, but I don’t think it’s arrogant to say what I did. You are basing your view of humanity on what you think humanity has done, and basing your view on AI based on what you think it will do. Those are fundamentally different and not comparable. If you want to talk about the science fiction future of AI, we should talk about the science fiction future of humanity as well. Let’s talk about augmenting ourselves, extending lifespans, and all of the good things that people think we’ll do in the coming centuries. If you want to look at humans and say that we haven’t evolved at all in the last 3000 years, then we should look at computers the same way. Computers haven’t “evolved” at all. They still do the same thing they always have. They do a lot more of it, but they don’t do anything “new”. We have found ways to increase the processing power, and the storage capacity, but a computer today has the same limits that the one that sent us to the moon had. It’s a computer, and incapable of original thought. You seem to believe that just because we throw more ram and processors at it that somehow that will change things, but it doesn’t. It just means we can do the same things, but faster. Eventually we’ll run out of things to process and data to store, but that won’t bring AI any closer to reality. We are climbing the mountain, but you speak like we have already crested. We’ve barely left base camp in the grand scheme of artificial intelligence.
Apart from your use of infinite I agree, there is no reason we shouldn’t be able to surpass nature with synthetic intelligence. The time computers have existed is a mere blip on a historic scale, and computers has surpassed us at logic games like Chess and at math already long ago.
Modern LLM models are just the current stage, before that it could be said it was pattern recognition. We had OCR in the 80’s as probably the most practical example. It may seem there is long between the breakthroughs, but 40 years is nothing compared to evolution.
I have no doubt strong AI will be achieved eventually, and when we do, I have no doubt AI will surpass our intelligence in every way very quickly.
As a counter argument against that, companies are trying to make self driving cars work for 20 years. Processing power has increased by a million and the things still get stuck. Pure processing power isn’t everything.
Everything is magic if you don’t understand how the thing works.
I wish. I don’t understand why my stomach can’t handle corn, but it doesn’t lead to magic. It leads to pain.
Have you eaten hominy corn? The nixtamalisation process makes it digestible.
I don’t have access to that, sadly. I’m pretty sure my body would reject it however. At least from my reading on what it is.
Sam Altman will make a big pile of investor money disappear before your very eyes.
The masses have been treating it like actual magic since the early stages and are only slowly warming up to the idea it‘s calculations. Calculations of things that are often more than the sum of it‘s parts as people start to realize. Well some people anyway.
If only.
I hope it collapses in a fire and we can just keep our foss local models with incremental improvements, that way both techbros and artbros eat shit
Unfortunately for that outcome, brute forcing with more compute is pretty helpful for now
As I often mention when this subject pops up: while the current statistics-based generative models might see some application, I believe that they’ll be eventually replaced by better models that are actually aware of what they’re generating, instead of simply reproducing patterns. With the current models being seen as “that cute 20s toy”.
In text generation (currently dominated by LLMs), for example, this means that the main “bulk” of the model would do three things:
- convert input tokens into sememes (units of meaning)
- perform logic operations with the sememes
- convert sememes back into tokens for the output
Because, as it stands, LLMs are only chaining tokens. They might do this in an incredibly complex way, but that’s it. That’s obvious when you look at what LLM-fuelled bots output as “hallucination” - they aren’t the result of some internal error, they’re simply an undesired product of a model that sometimes outputs desirable stuff too.
Sub “tokens” and “sememes” with “pixels” and “objects” and this probably holds true for image generating models, too. Probably.
Now, am I some sort of genius for noticing this? Probably not; I’m just some nobody with a chimp avatar, rambling in the Fediverse. Odds are that people behind those tech giants already noticed the same ages ago, and at least some of them reached the same conclusion - that better gen models need more awareness. If they are not doing this already, it means that this shit would be painfully expensive to implement, so the “better models” that I mentioned at the start will probably not appear too soon.
Most cracks will stay there; Google will hide them with an obnoxious band-aid, OpenAI will leave them in plain daylight, but the magic trick will still not be perfect, at least in the foreseeable future.
And some might say “use MOAR processing power!”, or “input MOAR training data!”, in the hopes that the current approach will “magically” fix itself. For those, imagine yourself trying to drain the Atlantic with a bucket: does it really matter if you use more buckets, or larger buckets? Brute-forcing problems only go so far.
Just my two cents.
I agree 100%, and I think Zuckerberg’s attempt at a massive 340,000 of Nvidia’s H100 GPUs AI based on LLM with the aim to create a generel AI sounds stupid. Unless there’s a lot more to their attempt, it’s doomed to fail.
I suppose the idea is something about achieving critical mass, but it’s pretty obvious, that that is far from the only factor missing to achieve general AI.
I still think it’s impressive what they can do with LLM. And it seems to be a pretty huge step forward. But It’s taken about 40 years from we had decent “pattern recognition” to get here, the next step could be another 40 years?
I think that Zuckerberg’s attempt is a mix of publicity stunt and “I want [you] to believe!”. Trying to reach AGI through a large enough LLM sounds silly, on the same level as “ants build, right? If we gather enough ants, they’ll build a skyscraper! Chrust me.”
In fact I wonder if the opposite direction wouldn’t be a bit more feasible - start with some extremely primitive AGI, then “teach” it Language (as a skill) and a language (like Mandarin or English or whatever).
I’m not sure on how many years it’ll take for an AGI to pop up. 100 years perhaps, but I’m just guessing.
That’s a huge oversimplification of the way LLMs work. They’re not statistical in the way a Markov chain is. They use neural networks, which are a decent analogy for the human brain. The way the synapses between neurons are wired is obviously different, and the way the neurons are triggered and the types of signals they can send to other neurons is obviously different. But overall, similar capabilities can in theory be achieved with either method. If you’re going to call neural networks statistics based, you might as well call the human brain statistics based as well.
That’s a huge oversimplification of the way LLMs work.
I’m sticking to what matters for the sake of the argument. Anyone who wants to inform themself further has a plethora of online resources to do so.
They’re not statistical in the way a Markov chain is.
Implied: “you’re suggesting that they work like Markov chains, they don’t.”
In no moment I mentioned or even implied Markov chains. My usage of the verb “to chain” is clearly vaguer within that context; please do not assume words onto my mouth.
They use neural networks, which are a decent analogy for the human brain. The way the synapses between neurons are wired is obviously different, and the way the neurons are triggered and the types of signals they can send to other neurons is obviously different. But overall, similar capabilities can in theory be achieved with either method.
I don’t disagree with the conclusion (i.e. I believe that neural networks can achieve human-like capabilities), but the argument itself is such a fallacious babble (false equivalence) that I’m not bothering further with your comment.
And it’s also an “ackshyually” given this context dammit. I’m not talking about the bloody neural network, but how it is used.
No need to get offended. Maybe I misunderstood the intent behind your original message. I think you made a lot of good points.
I brought up the Markov chain because a common misconception I’ve seen on the Internet and in real life is that LLMs work pretty much the same as Markov chains under the hood. And I saw no mention of neural networks in your original comment.
I found this graph very clear
Well, natural language processing is placed in the trough of disillusionment and projected to stay there for years. ChatGPT was released in November 2022…
Trying to make real and good use of AI generative models are cracks in the magic.
It’s pretty useful if you know exactly what you want and how to work within it’s limitations.
Coworkers around me already use ChatGPT to generate code snippets for Python, Excel VBA, etc. to good success.
Right, it’s a tool with quirks, techniques and skills to use just like any other tool. ChatGPT has definitely saved me time and on at least one occasion, kept me from missing a deadline that I probably would have missed if I went about it “the old way” lmao
You mean they’re using it to write boilerplate which shouldn’t have been written in the first place.
Why? If you know how to incorporate “boilerplate” and modify it correctly into your own code, what difference does it make if its from ChatGPT or Stackoverflow?
Difference to copy and paste from stackoverflow, probably not terribly much. The latter is already bad.
It’s as if the young’uns heard the term “10x developer” and decided that not understanding what you’re doing is the way to get there.
“This post is for paid subscribers”
(Also that page has a script I had to override just to copy and paste that)
It’s well worth reading the longer newsletter the above link quotes: https://www.wheresyoured.at/sam-altman-fried/
I kinda agree we are probably cresting the peak of the hype cycle right now.
