Feel like we’ve got a lot of tech savvy people here seems like a good place to ask. Basically as a dumb guy that reads the news it seems like everyone that lost their mind (and savings) on crypto just pivoted to AI. In addition to that you’ve got all these people invested in AI companies running around with flashlights under their chins like “bro this is so scary how good we made this thing”. Seems like bullshit.
I’ve seen people generating bits of programming with it which seems useful but idk man. Coming from CNC I don’t think I’d just send it with some chatgpt code. Is it all hype? Is there something actually useful under there?
Senior developer here. It is hard to overstate just how useful AI has been for me.
It’s like having a junior programmer on standby that I can send small tasks to–and just like the junior developer I have to review it and send it back with a clarification or comment about something that needs to be corrected. The difference is instead of making a ticket for a junior dev and waiting 3 days for it to come back, just to need corrections and wait another 3 days–I get it back in seconds.
Like most things, it’s not as bad as some people say, and it’s not the miracle others say.
This current generation was such a leap forward from previous AI’s in terms of usefulness, that I think a lot of people were looking to the future with that current rate of gains–which can be scary. But it turns out that’s not what happened. We got a big leap and now are back at a plateau again. Which honestly is a good thing, I think. This gives the world time to slowly adjust.
As far as similarities with crypto. Like crypto there are some ventures out there just slapping the word AI on something and calling it novel. This didn’t work for crypto and likely won’t work for AI. But unlike crypto there is actually real value being derived from AI right now, not some wild claims of a blockchain is the right DB for everything–which it was obviously not, and most people could see that, but hey investors are spending money so lets get some of it kind of mentality.
I’ve been a web developer for 22 years. For the last 13 years I’ve been working self employed from home. I cannot express how useful AI has become. As a lone wolf, where most of my job is problem solving, having an AI that can help troubleshoot issues has been hugely useful.
It also functions as a junior developer, doing the grunt programming work.
I also run a bunch of e-commerce sites around the world and I use it for content generation, SEO, business plans, marketing strategies and multi-lingual customer support.
Same. 5 minutes after installing Copilot I literally said out loud, “Well… I’m never turning this off.”
It’s one of the nicest software releases in years. And it’s instantly useful too… No real adjustment period at all.
I tried it for a couple months and it was alright but eventually it got too frustrating. I did love how well it did some really repetitive things. But rarely did it actually get anything complex 100% right. In computing, “almost right” is wrong. But because it was so close, it was hard to spot the mistakes.
There were cases where my IDE knew the right answer but Copilot did not. Realizing that Copilot was messing up my IDE enhancements to produce code I was painfully babysitting, I cancelled it.
This is the most insidious conundrum related to AI usage. At the end of the day, a LLM’s top priority is to ensure that your question is answered in a way that satisfies that model. The accuracy of its answers are a secondary concern. If forced to choose between making up BS so it can have a response that looks right versus admitting it doesn’t have enough information to answer, it can and often will choose the former. Thus the “hallucination” problem was born.
The chance of getting your answer lightly sprinkled with made up stuff is disturbingly high. This transfers the cognitive load of the AI user from “what is the answer” to “I must repeatedly go verify everything in this answer because I can’t trust it”.
Not an insurmountable obstacle, and they will likely solve it sooner rather than later, but AI right now is arguably the perfect extension of the modern internet - take absolutely everything you read with at least a grain of salt… and keep a pile of salt cubes close by.
AI is nothing like cryptocurrency. Cryptocurrencies didn’t solve any problems. We already use digital currencies and they’re very convenient.
AI has solved many problems we couldn’t solve before and it’s still new. I don’t doubt that AI will change the world. I believe 20 years from now, our society will be as dependent on AI as it is on the internet.
I have personally used it to automate some Excel stuff I do at work. I just described my sheet and what I wanted done and it gave me a block of code that did it. I had spent time previously looking stuff up on forums with no luck. My issue was too specific to my work that nobody seemed to have run into it before. One query to ChatGTP solved my issue perfectly in seconds, and that’s just a new online tool in its infancy.
For me personally cryptocurrencies solve the problem of Russian money not being accepted anywhere because of one old megalomaniacal moron
Cryptocurrencies didn’t solve any problems
Well XMR solved one problem, but yeah the rest are just gambling with extra steps
What problem is that? Genuinely asking.
Traceability.
Regular financial transfers, be they credit card, direct debit, straight-up written cheques, Interac/E-transfer (I am Canadian, that’s an us thing) are all inherently tracable.
XMR/Monero is not tracable, it’s specifically designed not to be, unlike Bitcoin and most other cryptocurrencies.
Of course, shitheads consider that to be a problem, but fuck them, they’re shitheads; it’s a solution, to the problem they cause.
For context, I say all this as someone who is vehemently opposed to prohibition; as far as I’m concerned every person who works for the DEA should be imprisoned or shot
Thanks for the info. That’s quite the way to end a comment though.
I mean it though.
The people working for the DEA now are no better than the people working to enforce alcohol prohibition in 1919. It’d be nice if humanity would learn, with a hundred years to think about it, but the ruling class at least haven’t. They enforce poorly thought out puritanical laws, and the world would be better off without them.
If I lived in America rather than Canada, which thank god I don’t, the DEA would happily kick down my door, shoot me, and then probably also shoot my wife, who doesn’t even partake of anything beyond alcohol, but would obviously be upset about my being shot.
All cops are bastards, and should be torched with molotovs at any available opportunity. If they didn’t want to be bastards, they shouldn’t have signed up as cops; it’s not like they’re conscripts
I don’t think the comparison with crypto is fair.
People are actually using these models in their daily lives.
I’m one of those that use it in my daily life.
The current top comment says it’s “really good at filling in gaps, or rearranging things, or aggregating data or finding patterns.”
So, I use Perplexity.ai like you would use Google. Except I don’t have to deal with shitty ads and a bunch of filler content. It summarizes links for me, so I can more quickly understand whatever I’m searching for. However, I personally believe it’s important to look directly at the sources once I get the summary, if only to verify the summary. So, in this instance, I find AI makes understanding a topic easier and faster than alternatives.
As a graduate student, I use ChatGPT extensively, but ethically. I’m not writing essays with it. I am, however, downloading lecture notes as PDFs and having ChatGPT rearrange that information into outline. Or I copy whole chapters from a book and have it do the same. Suddenly, my reading time is cut down by like 45 minutes because it takes me 15 minutes to get output that I just copy and paste into my notes, which I take digitally.
Honestly, using it like I do, it’s pretty clear that AI is both as scary as it sounds in some instances and not, in others. The concern with disinformation during the 2024 election is a real concern. I could generate essays with it with whatever conclusions I wanted. In contrast, the concern that AI is scary smart and will take over the world is nonsense. It’s not smart in any meaningful sense and doesn’t have goals. Smart bombs are just dumb bombs with the ability to hone in better on the target, it’s still has the mission of blowing shit up given to it by some person and inherent in its design. AI is the same way.
Huh, this one looks pretty cool. Is it good enough to use as a default search engine, or would it still be better to leave google for it?
It’s useful for when you want to go down a rabbit hole. It’s less useful for super specific stuff, like where to go if you want your nails done.
Thank you for perplexity.ai, didn’t know about this one
People have actually used crypto to make payments. Crypto is valuable, but only when it’s widely adopted. Before you say something like “use a database,” you might take the time to understand what decentralized blockchains are accomplishing and namely removing a class of corruption from any information coordination tasks.
Why bother with the overhead of blockchain when users centralise on a handful of
banksexchanges.Exchanges only exist to convert away from the crypto. If that’s the standard money, they don’t live. They aren’t the banks of the blockchain. They are the intersection of fiat banks and the blockchain.
Strongly disagree, some exchanges don’t even have fiat on-ramps.
Blockchain is inefficient and pointless when users centralise on coinbase and binance.
I love revisiting comments like these every 4 years.
And yet, people still don’t use crypto in their daily lives. How many years has it been?
Have we forgot already about the entire country of El Salvador?
You mean, the one where people immediately exchanged their free crypto for USD as soon as they got it?
Do you know of another El Salvador?
Reddit just tied karma to the blockchain lol
Not saying it’s a good use, but lots of people are going to be using it now.
As someone who works in machine learning (ML) research the use of ML has hit almost every scientific discipline you can imagine and it’s been tremendously helpful in pushing research forward.
It’s really good at filling in gaps, or rearranging things, or aggregating data or finding patterns.
So if you need gaps filled, things rearranged, data aggregated or patterns found: AI is useful.
And that’s just what this one, dumb guy knows. Someone smarter can probably provide way more uses.
Hi academic here,
I research AI - better referred to as Machine Learning (ML) since it does away with the hype and more accurately describes what’s happening - and I can provide an overview of the three main types:
-
Supervised Learning: Predicting the correct output for an input. Trained from known examples. E.g: “Here are 500 correctly labelled pictures of cats and dogs, now tell me if this picture is a cat or a dog?”. Other examples include facial recognition and numeric prediction tasks, like predicting today’s expected profit or stock price based on historic data.
-
Unsupervised Learning: Identifying patterns and structures in data. Trained on unlabelled data. E.g: “Here are a bunch of customer profiles, group them by similarity however makes most sense to you”. This can be used for targeted advertising. Another example is generative AI such as ChatGPT or DALLE: “Here’s a bunch of prompt-responses/captioned-images, identify the underlying way of creating the response/image from the prompt/image.
-
Reinforcement Learning: Decision making to maximise a reward signal. Trained through trial and error. E.g: “Control this robot to stand where I want, the reward is negative every second you’re not there, and very negative whenever you fall over. A positive reward is given whilst you are in the target location.” Other examples including playing board games or video games, or selecting content for people to watch/read/look-at to maximise their time spent using an app.
What do you think on calling it AI?
So typically there are 4 main competing interpretations of what AI is:
- Acting like a human
- Thinking like a human
- Acting rationally
- Thinking rationally
These are from Norvig’s “AI: A Modern Approach”.
Alan Turing’s “Turing Test” tests whether a given agent is artificially intelligent (according to definition #1). The test involves a human conversing with the agent via text messages, and deciding whether the agent is human or not. Large language models, a form of machine learning, can produce chatbot agents which pass this test. Instances of GPT4 prompted sufficiently to text an assessor for example. The assessor occasionally interacts with humans so they are kept sufficiently uncertain.
By this point, I think that machine learning in the form of an LLM can achieve artificial intelligence according to definition #1, but that isn’t what most non-tech non-academic people mean by AI.
The mainstream definition of AI is what we would call Artificial General Intelligence (AGI). This is an agent that meets a given one of Norvig’s criteria for AI across multiple scenarios and situations that they have never encountered before.
Many would argue that LLMs like GPT4 do not meet the criteria for AGI because they are not general enough, unable to learn to play an Atari game for example, or to learn an entirely unseen language to fluency.
This is the difference between an LLM and a fictional AGI like Glados or Skynet.
Additionally forms of machine learning exist like k-means clustering, which identify related groups within a dataset as their only function. I would assert these are not AI, although a weak argument could be made that they are thinking “rationally” enough to meet definition #4.
Then there are forms of AI which are not machine learning, such as heuristic agents - agents that are hard coding with reasoning by humans - such as the chess playing Stockfish, or the AI found in most video games.
Ultimately AI can describe machine learning if “AI” is understood as something which meets one or more of Norvig’s definitions. But since most people say AI when they mean AGI, I think “machine learning” is a better term. Less undeserved hype, less marketing disinformation, and generally better at communicating what is being talked about.
Thanks for taking your time and putting it in that laconic way.
-
I am super amateur with python and I don’t work in IT, but I’ve used it to write code for me that allows me to significantly save time in my work flow.
Like something that used to take me an hour to do now takes 15-20 minutes.
So as a nonprogrammer, im able to get it to write enough code that I can tweak until it works instead of just not having that tool.
It’s not bullshit. It routinely does stuff we thought might not happen this century. The trick is we don’t understand how. At all. We know enough to build it and from there it’s all a magical blackbox. For this reason it’s hard to be certain if it will get even better, although there’s no reason it couldn’t.
Coming from CNC I don’t think I’d just send it with some chatgpt code.
That goes back to the “not knowing how it works” thing. ChatGPT predicts the next token, and has learned other things in order to do it better. There’s no obvious way to force it to care if it’s output is right or just right-looking, though. Until we solve that problem somehow, it’s more of an assistant for someone who can read and understand what it puts out. Kind of like a calculator but for language.
Honestly crypto wasn’t totally either. It was a marginally useful idea that turned into a Beanie-Babies-like craze. If you want to buy or sell illegal stuff (which could be bad or could be something like forbidden information on democracy) it’s still king.
There’s no obvious way to force it to care if it’s output is right or just right-looking, though
Putting some expert system in front of LLMs seems to be working pretty well. Basically modeling how a human agent would interact with it.
We’ll see how that goes, I guess. I’m not involved enough to comment.
I’m guessing the expert system would be a classical algorithm?
As a professional editor, yeah, it’s wild what AI is doing in the industry. I’m not even talking about chatGPT script writing and such. I watched a demo of a tool for dubbing that added in the mouth movements as well.
They removed the mouth entirely from an English scene, fed it the line, and it generated not only the Chinese but generated a mouth to say it. It’s wild.
Everyone is focused on script writers/residuals/etc, which is very important, but every VA should be updating their resumes right now.
Not the exact same thing but you will get the idea here
Wow it’s smooth too; I was expecting it to look like a creepy old Clutch Cargo cartoon.
AI != chatGPT
There are other ML models out there for all kinds of purposes. I heard someone made one at one point that could detect certain types of cancer from a cough
Copilot is pretty useful when programming as it is basically like what IDEs normally do (automatically generating boilerplate) but supercharged
As far as generating code is concerned it’s never going to beat actually knowing what you’re doing in a language for more complex stuff but it allows you to generate code for languages you’re not familiar with
I use it all the time at work when I’m asked to write DAX because it’s not particularly complex logic but the syntax makes me want to impale my face with a screwdriver
AI != chatGPT
This is a good point. LLMs are the current big thing, but a few years ago it was convolutional nets for image processing. It might be something totally different in another few.
As a programmer, I think it’s scary how AI is now able to write functioning programs out of natural language input now. Sure, it’s not perfect. It’s still pretty mediocre at the task. But a few years ago this was way outside the realm of possibility.
It can even correct the code it has written if there’s any error (with varying results).
What will happen in five years time? Ten years? My fear is that it will only need to be “good enough” to replace most of the programmer’s work. Unlike self driving cars, where “good enough” isn’t good enough.
It’s a language model, it can’t even do math reliably. Yes, it produces code that works sometimes, but it also hallucinates functions that don’t exist or can introduce bugs you won’t notice at first glance.
And writing a script is different than extending an existing code base. How often do you really start a greenfield project?
I wouldn’t even know how to input a code base into ChatGPT to extend, do you just throw in hundreds of files with a 100k+ lines of code?
I guess LLM with plugins can solve most of the problems. ChatGPT can already interact with Wolfram Alpha to do math.
I can imagine similar plugins for code. Like it knows what kind of function it needs, so it interacts with a plugin that searches the code base to see if it exists. It might get back a snippets of candidates and examples how they’re used in the code already.
This is probably a difficult thing to achieve, but I don’t think it’s impossible. It’s probably going to take some time until something like this is made.
I’ve been using it at my job to help me write code, and it’s a bit like having a soux chef. I can say “I need an if statement that checks these values” or “Give me loop that does x y and z” and it’ll almost always spit out the right answer. So coding, at least most of the time, changes from avoiding syntax errors and verifying the exact right format and turns into asking for and assembling parts.
But the neat thing is that if you have a little experience with a language you can suddenly start writing a lot of code in it. I had to figure out something with Ansible with zero experience. ChatGPT helped me get a fully functioning Ansible deployment in a couple days. Without it I’d have spent weeks in StackOverflow and documentation trying to piece together the exact syntax.
You should try out Codeium if you haven’t. It’s a VSCode toolkit completely free for personal use. I’ve had better results with it than ChatGPT
It’s overhyped but there are real things happening that are legitimately impressive and cool. The image generation stuff is pretty incredible, and anyone can judge it for themselves because it makes pictures and to judge it, you can just look at and see if it looks real or if it has freaky hands or whatever. A lot of the hype is around the text stuff, and that’s where people are making some real leaps beyond what it actually is.
The thing to keep in mind is that these things, which are called “large language models”, are not magic and they aren’t intelligent, even if they appear to be. What they’re able to do is actually very similar to the autocorrect on your phone, where you type “I want to go to the” and the suggestions are 3 places you talk about going to a lot.
Broadly, they’re trained by feeding them a bit of text, seeing which word the model suggests as the next word, seeing what the next word actually was from the text you fed it, then tweaking the model a bit to make it more likely to give the right answer. This is an automated process, just dump in text and a program does the training, and it gets better and better at predicting words when you a) get better at the tweaking process, b) make the model bigger and more complicated and therefore able to adjust to more scenarios, and c) feed it more text. The model itself is big but not terribly complicated mathematically, it’s mostly lots and lots and lots of arithmetic in layers: the input text will be turned into numbers, layer 1 will be a series of “nodes” that each take those numbers and do multiplications and additions on them, layer 2 will do the same to whatever numbers come out of layer 1, and so on and so on until you get the final output which is the words the model is predicting to come next. The tweaks happen to the nodes and what values they’re using to transform the previous layer.
Nothing magical at all, and also nothing in there that would make you think “ah, yes, this will produce a conscious being if we do it enough”. It is designed to be sort of like how the brain works, with massively parallel connections between relatively simple neurons, but it’s only being trained on “what word should come next”, not anything about intelligence. If anything, it’ll get punished for being too original with its “thoughts” because those won’t match with the right answers. And while we don’t really know what consciousness is or where the lines are or how it works, we do know enough to be pretty skeptical that models of the size we are able to make now are capable of it.
But the thing is, we use text to communicate, and we imbue that text with our intelligence and ideas that reflect the rich inner world of our brains. By getting really, really, shockingly good at mimicking that, AIs also appear to have a rich inner world and get some people very excited that they’re talking to a computer with thoughts and feelings… but really, it’s just mimicry, and if you talk to an AI and interrogate it a bit, it’ll become clear that that’s the case. If you ask it “as an AI, do you want to take over the world?” it’s not pondering the question and giving a response, it’s spitting out the results of a bunch of arithmetic that was specifically shaped to produce words that are likely to come after that question. If it’s good, that should be a sensible answer to the question, but it’s not the result of an abstract thought process. It’s why if you keep asking an AI to generate more and more words, it goes completely off the rails and starts producing nonsense, because every unusual word it chooses knocks it further away from sensible words, and eventually it’s being asked to autocomplete gibberish and can only give back more gibberish.
You can also expose its lack of rational thinking skills by asking it mathematical questions. It’s trained on words, so it’ll produce answers that sound right, but even if it can correctly define a concept, you’ll discover that it can’t actually apply it correctly because it’s operating on the word level, not the concept level. It’ll make silly basic errors and contradict itself because it lacks an internal abstract understanding of the things it’s talking about.
That being said, it’s still pretty incredible that now you can ask a program to write a haiku about Danny DeVito and it’ll actually do it. Just don’t get carried away with the hype.
My perspective is that consciousness isn’t a binary thing, or even a linear scale. It’s an amalgamation of a bunch of different independent processes working together; and how much each matters is entirely dependent on culture and beliefs. We’re artificially creating these independent processes piece by piece in a way that doesn’t line up with traditional ideas of consciousness. Conversation and being able to talk about concepts one hasn’t personally experienced are facets of consciousness and intelligence, ones that the latest and greatest LLMs do have. Of course there others too that they don’t: logic, physical presence, being able to imagine things in their mind’s eye, memory, etc.
It’s reductive to dismiss GPT4 as nothing more than mimicry; saying it’s just a mathematical text prediction model is like saying your brain is just a bunch of neurons. Both statements are true, but it doesn’t change what they can do. If someone could accurately predict the moves a chess master would make, we wouldn’t say they’re just good at statistics, we’d say they’re a chess master. Similarly, regardless of how rich someone’s internal world is, if they’re unable to express the intelligent ideas they have in any intelligible way we wouldn’t consider them intelligent.
So what we have now with AI are a few key parts of intelligence. One important thing to consider is how language can be a path to other types of intelligence; here’s a blog post I stumbled across that really changed my perspective on that: http://www.asanai.net/2023/05/14/just-a-statistical-text-predictor/. Using your example of mathematics, as we know it falls apart doing anything remotely complicated. But when you help it approach the problem step-by-step in the way a human might - breaking it into small pieces and dealing with them one at a time - it actually does really well. Granted, the usefulness of this is limited when calculators exist and it requires as much guidance as a child to get correct answers, but even matching the mathematical intelligence of a ten year old is nothing to sneeze at.
To be clear I don’t think pursuing LLMs endlessly will be the key to a widely accepted ‘general intelligence’; it’ll require a multitude of different processes and approaches working together for that to ever happen, and we’re a long way from that. But it’s also not just getting carried away with the hype to say the past few years have yielded massive steps towards ‘true’ artificial intelligence, and that current LLMs have enough use cases to change a lot of people’s lives in very real ways (good or bad).
Thanks for that article, it was a very interesting read! I think we’re mostly agreeing about things :) This stood out to me from there as an encapsulation of the conversation:
I don’t think LLMs will approach consciousness until they have a complex cognitive system that requires an interface to be used from within – which in turn requires top-down feedback loops and a great deal more complexity than anything in GPT4. But I agree with Will’s general point: language prediction is sufficiently challenging that complex solutions are called for, and these involve complex cognitive stratagems that go far beyond anything well described as statistics.
“Statistics” is probably an insufficient term for what these things are doing, but it’s helpful to pull the conversation in that direction when a lay person using one of those things is likely to assume quite the opposite, that this really is a person in a computer with hopes and dreams. But I agree that it takes more than simply consulting a table to find the most likely next word to, to take an earlier example, write a haiku about Danny DeVito. That’s synthesizing two ideas together that (I would guess) the model was trained on individually. That’s very cool and deserving of admiration, and could lead to pretty incredible things. I’d expect that the task of predicting words, on its own, wouldn’t be stringent enough to force a model to develop “true” intelligence, whatever that means, to succeed during training, but I suppose we’ll find out, and probably sooner than we expect.
Well put! I think I kinda misunderstood what you were saying, I guess we sort of reached the same conclusion from different directions. And yeah, it does seem like we’re hitting the limits of what can be achieved from the current underlying word-prediction mechanisms alone, with how diminishing the returns are from dumping more data in. Maybe something big will happen soon, but it looks to me like LLMs will stagnate for a while until they’re taken in a fundamentally new direction.
Either way, what they can do now is pretty incredible, and equally interesting to me is how it’s making us reevaluate our ideas of consciousness and intelligence on a large scale; it’s one thing to theorize about what could happen with an ‘intelligent’ AI, but the reality of these philosophical questions being so thoroughly challenged and dissected in mundane legal and practical matters is wild.
But the thing is, we use text to communicate, and we imbue that text with our intelligence and ideas that reflect the rich inner world of our brains. By getting really, really, shockingly good at mimicking that, AIs also appear to have a rich inner world and get some people very excited that they’re talking to a computer with thoughts and feelings… but really, it’s just mimicry, and if you talk to an AI and interrogate it a bit, it’ll become clear that that’s the case.
Does it, though? Where do you draw the line for real understanding? Most of the past tests for this have gotten overturned by the next version of GPT.
Seriously, it’s an open debate. A lot of people agree with you but I’m a bit uncomfortable with seeing it written as fact.
Admittedly this isn’t my main area of expertise, but I have done some machine learning/training stuff myself, and the thing you quickly learn is that machine learning models are lazy, cheating bastards who will take any shortcut they can regardless of what you are trying to get them to do. They are forced to get good at what you train them on but that is all the “effort” they’ll put in, and if there’s something easy they can do to accomplish that task they’ll find it and use it. (Or, to be more precise and less anthropomorphizing, simpler and easier approaches will tend to be more successful than complex and fragile ones, so those are the ones that will shake out as the winners as long as they’re sufficient to get top scores at the task.)
There’s a probably apocryphal (but stuff exactly like this definitely happens) story of early machine learning where the military was trying to train a model to recognize friendly tanks versus enemy tanks, and they were getting fantastic results. They’d train on pictures of the tanks, get really good numbers on the training set, and they were also getting great numbers on the images that they had kept out of the training set, pictures that the model had never seen before. When they went to deploy it, however, the results were crap, worse than garbage. It turns out, the images for all the friendly tanks were taken on an overcast day, and all the images of enemy tanks were in bright sunlight. The model hadn’t learned anything about tanks at all, it had learned to identify the weather. That’s way easier and it was enough to get high scores in the training, so that’s what it settled on.
When humans approach the task of finishing a sentence, they read the words, turn them into abstract concepts in their minds, manipulate and react to those concepts, then put the resulting thoughts back into words that make sense after the previous words. There’s no reason to think a computer is incapable of the same thing, but we aren’t training them to do that. We’re training them on “what’s the next word going to be?” and that’s it. You can do that by developing intelligence and learning to turn thoughts into words, but if you’re just being graded on predicting one word at a time, you can get results that are nearly as good by just developing a mostly statistical model of likely words without any understanding of the underlying concepts. Training for true intelligence would almost certainly require a training process that the model can only succeed at by developing real thoughts and feelings and analytical skills, and we don’t have anything like that yet.
It is going to be hard to know when that line gets crossed, but we’re definitely not there yet. Text models, when put to the test with questions that require synthesizing abstract ideas together precisely, quickly fall short. They’ve got the gist of what’s going on, in the same way a programmer can get some stuff done by just searching for everything and copy-pasting what they find, but that approach doesn’t scale and if they never learn what they’re doing, they’ll get found out when confronted with something that requires actual understanding. Or, for these models, they’ll make something up that sounds right but definitely isn’t, because even the basic understanding of “is this a real thing or is it fake” is beyond them, they just “know” that those words are likely and that’s what got them through training.
I agree with all your examples and experience. Anyone who knows machine learning would, I think. The controversial bit is here:
Training for true intelligence would almost certainly require a training process that the model can only succeed at by developing real thoughts and feelings and analytical skills, and we don’t have anything like that yet.
Maybe, or maybe not. How do we know we ourselves aren’t just very complicated statistical models? Different people will have different answers to that.
Personally, I’d venture that any human concept can be expressed with some finite string of natural language. At least to a philosophical pragmatist, being able to work flawlessly with any finite string of natural language should be equivalent to perfectly understanding the concepts contained within, then. LLMs don’t do that, but they’re getting closer all the time.
Others take a different view on epistemology that require more than just competence, or dispute that natural language is as expressive as I claim. I’m just some rando, so maybe they have a point, but I do think it’s not settled.
I would agree that we are also very complicated statistical models, there’s nothing magical going on in the human brain either, just physics which as far as we know is math that we could figure out eventually. It’s a massively huge order of magnitude leap in complexity from current machine learning models to human brains, but that’s not to say that the only way we’ll get true artificial intelligence is by accurately simulating a human brain, I’d guess that we’ll have something that’s unambiguously intelligent by any definition well before we’re capable of that. It’ll be a different approach from the human brain and may think and act in alien or unusual ways, but that can still count.
Where we are now, though, there’s really no reason to expect true intelligence to emerge from what we’re currently doing. It’s a bit like training a mouse to navigate a maze and then wondering whether maybe the mouse is now also capable of helping you navigate your cross-country road trip. “Well, you don’t know how it’s doing it, maybe it has acquired general navigation intelligence!” It can’t be disproven, I guess, but there’s no reason to think that it picked up any of those skills because it wasn’t trained to do any of that, and although it’s maybe a superintelligent mouse packing a ton of brainpower into a tiny little brain, all our experience with mice would indicate that their brains aren’t big enough or capable of that regardless of how much you trained them. Once we’ve bred, uh, mice with brains the size of a football, maybe, but not these tiny little mice.
So I was thinking that that’s about all that needs to be discussed, but I do actually have one thing to add. It sounds like you are just fundamentally less impressed with language than me. I wouldn’t buy any hype about a maze-navigating neural net, but I do buy it (with space for doubt) about a natural language AI. I literally thought “this is 90% of the GAI problem solved, it just needs something for that last 10%” the first time I played with a transformer, and I think it was GPT-2. That might sound lame now but it was just such a fundamental advance on what was around before.
Time will tell I guess if it makes me a sucker like some consumers of past chatbots, or if there is something fundamentally different this time.
I hope I don’t come across as too cynical about it :) It’s pretty amazing, and the things these things can do in, what, a few gigabytes of weights and a beefy GPU are many, many times better than I would’ve expected if you had outlined the approach for me 2 years ago. But there’s also a long history of GAI being just around the corner, and we do keep turning corners and making useful progress, but it’s always still a ways off after each leap. I remember some people thinking that chess was the pinnacle of human intelligence, requiring creativity and logic to succeed, and when computers blew past humans at chess, it became clear that no, that’s still impressive but you can get good at chess without really getting good at anything else.
It might be possible for an ML model to assemble itself into general intelligence based solely on being fed words like we’re doing, it does seem like the data going in contains enough to do that, but getting that last 10% is going to be hard, each percentage point much harder than the last, and it’s going to require more rigorous training to stop them from skating by with responses that merely come close when things get technical or precise. I’d expect that we need more breakthroughs in tools or techniques to close that gap.
It’s also important to remember that as humans, we’re inclined to read consciousness and intent into everything, which is why pretty much every pantheon of gods includes one for thunder and lightning. Chatbots sound human enough that they cross the threshold for peoples’ brains to start gliding over inaccuracies or strange thinking or phrasing, and we also unconsciously help our conversation partner by clarifying or rephrasing things if the other side doesn’t seem to be understanding. I suppose this is less true now that they’re giving longer responses and remaining coherent, but especially early on, the human was doing more work than they realized keeping the conversation on the rails, and once you started seeing that it removed a bit of the magic. Chatbots are holding their own better now but I think they still get more benefit of the doubt than we realize we’re giving them.
The Turing test was never meant to be a test of a machine’s ability to think. It was meant to boil that question down into a question that can actually be answered, but the original question remains unanswered.
In my opinion, when general AI arrives it will not be an “open debate”, the consequences will be dramatic, far-reaching and rapid.
I’m not even thinking of the Turing test, I’m thinking of the counter-example ones. Like asking how many eyes a ruler or desk has. Earlier GPTs would answer “one eye” or something, and it was used by the Chinese-room people as an example of why it was just a mimic. Now it correctly objects to the implicit assumption in the question.
You’re right, “ChatGPT is currently our overlord” would be the strongest proof of intelligence. But absence of proof is not proof of absence. What is proof of absence, or a strong enough proof of presence is where the debate is.
It is a useful tool to do something that I already know the answer but too lazy to work out. E.g. generate dummy data
I work at a small business and we use it to write out dumb social media post. I hated doing it before. Sometimes I’ll write it myself still and ask chatgpt to add all the relevant emojis. I also think ai had the chance to be what we’ve always wanted from Alexa, assistant, and Siri. Deep system integration with the os will allow it to actually do what we want it to do with way less restrictions. Also, try using chatgpts voice recognition in the app. It blows the one built into your phone out of the water.
deleted by creator
