I enjoy following your blog but it generally feels like you are cynical and would not change your mind even if given ample evidence that undermines your views. Also, just for fun, here is ChatGPT's reply to your blog post: Here are four counterarguments to the points made in the blog post:
1. **Struggles in Early Adoption Do Not Equate to Long-Term Failure**:
- The fact that many companies are struggling to deploy generative AI is not uncommon for any transformative technology in its early stages. Remember, the early days of the internet, cloud computing, and even e-commerce faced similar adoption hurdles. Challenges around cost and confusion can be temporary and often decrease as the technology matures and becomes more widely understood and accessible.
2. **Backlash and Criticism Can Lead to Improvement**:
- Every groundbreaking technology faces criticism. However, it's important to differentiate between constructive criticism, which can lead to improvement and iteration, and general skepticism. Moreover, linking AI’s future to a few negative headlines might be myopic. Just as ChatGPT and similar models have their detractors, they also have a vast number of supporters and users who find value in them.
3. **Missteps and Controversies Do Not Undermine the Entire Potential of AI**:
- The issue regarding Google’s LLM pointing out controversial figures as "greatest" leaders is a flaw, but it's crucial to separate the limitations of one model from the vast potential of the technology as a whole. AI models can and will be improved over time, and the emphasis should be on progress and refinement.
4. **Legal Issues and Economic Challenges are Part of Tech Evolution**:
- Many transformative technologies face legal challenges, especially in their early stages. This isn’t unique to AI. These challenges can lead to improved guidelines and practices for the industry. Furthermore, the mention of potential lawsuits is speculative. Even if OpenAI faces challenges, this does not mean that the entire field of generative AI will be rendered obsolete.
Lastly, on a broader note, technology's real value is often realized in the long run. Immediate setbacks or challenges do not necessarily predict a technology's long-term viability or success.
and: "Dismissing these arguments as merely saying “things will get better with time” neglects the historical context and specific challenges each technological advancement has overcome"
Also a specific example is that everyone is ok with criticising google's ai for listing Hitler, Stalin and Mussolini as great leaders, but few people (human intelligences they supposedly are) are willing to make the trivial observation that Stalin was indeed one of the greatest leaders of all time and that Hitler was a great leader in some respects (he was able to motivate the Germans well, he built a good power structure and he managed to conquer most of Europe [temporarily], also one may argue that being an evil villain doesn't preclude someone from being considered a great leader).
So people are unable to deconstruct a simple case, but we expect them to understand (critically and systemically) the big picture...
The issue here is that everyone f**ing ignores specific challenges. I came to this post from a post at LW ( https://www.lesswrong.com/posts/h6pFK8tw3oKZMppuC/is-this-the-beginning-of-the-end-for-llms-as-the-royal-road ) by my favourite Bill Benzon, who said "if we are to move to a level of accomplishment beyond what has been exhibited to date, we must understand what these engines are doing so that we may gain control over them. We must think about the nature of language and of the mind."
This is similar BTW to David Deutsch's 'Expecting to create an AGI without first understanding how it works is like expecting skyscrapers to fly if we build them tall enough.' (2012).
People who have some idea about our minds and language sometimes point out how understanding intelligence may be a prerequisite to building artificial intelligence.
But rarely people actually discuss specific challenges well enough. Gary doesn't do that, Rodney Brooks rarely does that, CEOs of AI companies (OpenAI, DeepMind, etc.) never do that, VCs don't do that, AI developers rarely do that, etc.
They aren't my arguments. They are ChatGPT's. Here is the rebuttal to your response: "Beyond just optimism for the future, these arguments draw from past technological trends to demonstrate that initial hurdles often lead to refinement, adaptation, and broader societal acceptance"
His point is not that the tech is worthless, but that he was an early skeptic of a technology that was being hailed as genuinely revolutionary and the cusp of "true" AI earlier this year.
Gary is not an AI cynic - he is on record as an advocate for AI as a valuable technology - his point is that ChatGPT and LLMs in particular might be turning into a case of "these aren't the droids you're looking for" more rapidly than even he expected.
I think what he is cynical about (and rightly so) is the hailing of a provably unreliable technology as a panacea for all sorts of complex challenges in inappropriate high-stakes problem domains (medicine, law etc.) - as well as the dubious wisdom of unleashing bullshit generators on a public already suffering from the brain-rot of the last unregulated technology revolution (internet, social media et al.) that was badly fumbled in the blinkered pursuit of techno-utopianism.
Quick Edit: Speaking of the last tech revolution and unbounded optimism, all of the problems we see today with the regulation of misinformation and the degradation of public discourse were all supposed to be solved by now (ironically by Machine Learning in many cases) - but have turned out to be largely intractable, as evidenced by the ongoing promulgation of lies and propaganda on the most powerful platforms and tech of our time (Google, Facebook etc.). Contrary to the once popular song, it isn't true that "things can only get better".
"Lastly, on a broader note, technology's real value is often realized in the long run. Immediate setbacks or challenges do not necessarily predict a technology's long-term viability or success."
The thing is, we already are in the long run. People think this all started with chat GPT, but before that was GPT 3, GPT 2, GPT 1, Machine learning, neural network models. ChatGPT is just the latest iteration of decades old technology.
I think much of the overhype is due to this false impression, that this is just the start of this technological implementation. But the reality, as far as I see it, is chatGPT is the pinnacle of what we can achieve with neural network based associative learning, given billions of dollars and decades of refining the implementation, and the world entire internet as a training database. ChatGPT appears to be the pinnacle of this approach given the world's resources and decades of development.
This so reminds me of blockchain's hype curve within the enterprise space. It took about 2-3 years before tech folks came to a consensus it was just another kind of database, one that was very hard to connect to other databases, particularly transactional, whether trusted or not.
One should not fall from the peak of hype to the depths of disappointment.
Architecturally, LLM are a big step forward. There are many areas that have large and repetitive data, but which is not organized in a way that a machine can handle. Here LLM will shine.
The ability to make use of third-party tools is also something very promising, and can be learned simply from examples, which are way cheaper to make than hiring top AI talent to build custom fancy models. Many of the failures of not being able to call tools correctly stems from too few examples.
In short, the long-term future is looking good, but a mid-course correction is likely expected.
My company is building things that no one would ever pay humans to do because it would be too costly. This is not creating NEW jobs, but it isn't taking any away. Maybe we're in a unique industry and every other industry will ONLY replace jobs with AI, but I suspect that a lot of formerly-impossible tasks will suddenly be unlocked.
And also, there probably will be mass unemployment relatively soon due to this stuff, at least in some industries. But I'm not prepared to say there is only one outcome and it's purely negative... yet.
Maybe the political process, or burning down some government builds to get a guaranteed mincome or to regulate all the AI companies to irrelevance might solve the problem. It will be a while before robots will be able to replace human hands and physical efforts, so maybe everyone just goes back to farming and carpentry?
I know people who have been made irrelevant in previous economic shifts, e.g. graphic design not being a real career anymore. It's really bad. I'm not celebrating it, but now you can go out and get a logo for free and start a lifestyle brand centered on how cool your group of influencer friends is, so the invisible hand taketh and giveth :P
I think fewer people with Bullshit Jobs (https://en.wikipedia.org/wiki/Bullshit_Jobs) might end up being a good thing at some point in the future. AI will definitely replace a lot of bullshit office jobs that make the workers miserable.
Can’t wait for my two options in life to be manual labor (while it lasts) or the faint hope of sufficient government welfare. Oh, or a violent revolution that may or may not yield anything other than further destabilization.
But honestly, I’m glad people are pouring billions into generative AI. Everything I listed above is made worth it by knowing that any creepy dude can take pictures off my old Facebook account and make deepfake porn of me.
I admit I've always been a bit miffed at how excited companies are over AI. Like, I think AI's really cool, but in terms of immediate economic applications? I dunno. Generative AI makes more sense than spending a hundred million to make a Starcraft AI (sorry DeepMind I love RL I just don't get why google paid for that) but still seems overhyped.
Still, I'm gonna bet that OpenAI specifically will do fine. A few reasons why:
1) Six months since GPT-4 and the only one who's close is Anthropic. And with the success of GPT-4 OpenAI will see a boost in funding and ability to attract talent, meaning there's every reason to think they will stick the landing of the eventual GPT-5.
2) The potential for coding is huge. GPT 4 is genuinely useful for coding. If there's one thing we should be confident that will come out of this LLM hype, it's bigger and better LLMs. So even if GPT-4 doesn't quite give companies their money's worth as of yet, the fact that they're getting experience with these systems and something even better is around the corner should still justify their investments.
3) Although some problems with LLMs will be hard to remove (hallucinations) others are much more fixable (annoying AI-speak). I think there'll be a new wave of excitement when LLMs that are comparable in power to GPT-4 but are actually RLHF'd to write well are released.
OpenAI is okay for coding in my experience. For me they're better at getting the skeleton of code down, or filling in boilerplate. E.g. "please generate a golang struct for the following JSON payload" GitHub's Copilot is also okay. It gets some things magically correct, but requires enough double checking that I can't let a single line get to production without testing it, though thankfully I'm a test driven development guy so this is not really a problem.
"You have NO idea what you are talking about" - applies to ChatGPT and friends, more than it does to people :)
For humans, 'model of the world' comes from... the world! For LLMs, it comes from... words (and other data). Words aren't substitutes for the world, except symbolically.
You're implying that education doesn't need to involve books, photographs, art and movies - or teachers. We just need to move around a lot and do lots of different things. But I don't see folks ripping up their precious attainment certificates anytime soon...
Boy there is a lot of anger in the comments about infidels daring to question our lord and savior LLMs. This really reminds me of blockchain and self-driving cars insanity. At this point susceptibility to corporate hype should qualify you for a disability.
I'm interested in this comment of yours, Gary: "Large Language Models aren’t like classical databases in which individual pieces of data can be removed at will; if any content is removed, the entire model must (so far as I understand it) be retrained, at great expense." Indeed.
Consider this remark from Fodor and Pylyshyn 1988 (Connectionism and cognitive architecture: A critical analysis):
"Classical theories are able to accommodate these sorts of considerations because they assume architectures in which there is a functional distinction between memory and program. In a system such as a Turing machine, where the length of the tape is not fixed in advance, changes in the amount of available memory can be affected without changing the computational structure of the machine; viz by making more tape available. By contrast, in a finite state automaton or a Connectionist machine, adding to the memory (e.g. by adding units to a network) alters the connectivity relations among nodes and thus does affect the machine’s computational structure. Connectionist cognitive architectures cannot, by their very nature, support an expandable memory, so they cannot support productive cognitive capacities. The long and short is that if productivity arguments are sound, then they show that the architecture of the mind can’t be Connectionist. Connectionists have, by and large, acknowledged this; so they are forced to reject productivity arguments."
It seems to me that that, the irreducible interweaving of memory and program in connectionist architecture, is a severe limitation. While Fodor and Pylyshyn are talking about adding items to the system it should be clear that excising them is problematic for the same reason. Such a system can neither learn nor forget, and that is true no matter how many parameters it has nor how large the corpus it is trained on.
I use LLMs some now. Will I use them more or less five years from now? I guess more.
The development cycle is similar to the internet/browser cycle, but faster Pets.Com, the crash, and over 10 years and with the hard work of engineers, the internet is embedded in daily activity. This may be 6 or 8 years to widespread usage.
It’s an old line. In 3 years it seems like less than you expected has happened. In 10 years more than you expected has happened.
Why do you guess “more”? Are you confident the hallucination problem will be solved? Additionally, what do you think will begin to happen to the major LLMs when the training data quality inevitably declines? The current iterations of LLMs are trained on just under 20 years of high-quality, human-produced internet content, but as time goes on, the internet gets more and more polluted with nonsensical AI-generated mush (I really started to notice it a few years ago). Additionally, since a lot of the training data may have been obtained illegally, a lot of high quality data will most likely need to be removed. I’m not so sure they’ll be able to replace it all with more high quality, usable data.
Overall, I get what you’re saying about how technology use/implementation increases over time (sometimes slowly), but I think LLMs have some pretty severe limitations that might spell doom for them
When I listened to the Republican Debate, Christie's jab towards Ramaswamy in regards to sounding like "ChatGPT" was one of the stand out moments for me.
I don't know if he came up with it beforehand or if it was thought of on the spot--the pure and distilled genius of it makes me think he thought of it contemporaneously during the debate. What I think he was trying to convey with this message is that Ramaswamy's positions / talking points were intellectually valid but paper thin and lacking meaning. Which I don't necessarily agree with but it was an incredible jab and too good to ignore.
Honestly, it shocks me how bad Siri is at transcribing my voice into written words. It guesses badly from context and enunciation makes things worse. I still think about how bad AI is in its most mainstreamed use cases. That is enough to temper my expectations
Or rather. It’s shocking how bad it is given how good it is. It is as if you had a power drill that works great on plywood that suddenly breaks when applied to cardboard.
I do think that even great technology can be undone by undermining a user’s sense of comfort. Something can be so good that its fail points become unacceptable because they are so hard to mentally model and predict given its successes. It’s sort of a paradox of usability because the tech will lead you astray by irresistably drawing you into its technical limits
Haha. This piece is both brutal and hilarious. I don't know whether to laugh or to feel sorry for those who have placed their faith in LLMs. Great read.
We should not be too excessive with the contempt towards Chat-GPT or similar tools. Because of the race for novelty, AI driven chat-bots were released too soon. But these first generation systems are performing surprisingly well for some situations. Because of the flaws and public opinion caution, the release of the next generation will be possibly much delayed but this second generation will be certainly much improved and an average user will not probably be able to detect any important errors. And this is when the use of AI tools by a large audience will become a very sever issue.
I don’t think the present temporary fall of chat-bots, if confirmed, will really slow down the very fast, much too fast as concerns societal impacts, global development of AI. I think that the main stream AI products are not for ordinary people. If the big tech companies will make less money with the general audience, they will focus and make their business with state agencies, organizations, banks, manufacturers. Governments and global companies will invest massively in AI applications because they will consider it as strategic and mandatory and this will give effects like in any enterprise where there is really a lot of money. Maybe even the today’s all public Chat-GPT is only a pale copy, a sub-product of a more reliable system intended for institutional or business customers.
I enjoy following your blog but it generally feels like you are cynical and would not change your mind even if given ample evidence that undermines your views. Also, just for fun, here is ChatGPT's reply to your blog post: Here are four counterarguments to the points made in the blog post:
1. **Struggles in Early Adoption Do Not Equate to Long-Term Failure**:
- The fact that many companies are struggling to deploy generative AI is not uncommon for any transformative technology in its early stages. Remember, the early days of the internet, cloud computing, and even e-commerce faced similar adoption hurdles. Challenges around cost and confusion can be temporary and often decrease as the technology matures and becomes more widely understood and accessible.
2. **Backlash and Criticism Can Lead to Improvement**:
- Every groundbreaking technology faces criticism. However, it's important to differentiate between constructive criticism, which can lead to improvement and iteration, and general skepticism. Moreover, linking AI’s future to a few negative headlines might be myopic. Just as ChatGPT and similar models have their detractors, they also have a vast number of supporters and users who find value in them.
3. **Missteps and Controversies Do Not Undermine the Entire Potential of AI**:
- The issue regarding Google’s LLM pointing out controversial figures as "greatest" leaders is a flaw, but it's crucial to separate the limitations of one model from the vast potential of the technology as a whole. AI models can and will be improved over time, and the emphasis should be on progress and refinement.
4. **Legal Issues and Economic Challenges are Part of Tech Evolution**:
- Many transformative technologies face legal challenges, especially in their early stages. This isn’t unique to AI. These challenges can lead to improved guidelines and practices for the industry. Furthermore, the mention of potential lawsuits is speculative. Even if OpenAI faces challenges, this does not mean that the entire field of generative AI will be rendered obsolete.
Lastly, on a broader note, technology's real value is often realized in the long run. Immediate setbacks or challenges do not necessarily predict a technology's long-term viability or success.
All of your counterarguments can be boiled down to "it will get better with time".
and: "Dismissing these arguments as merely saying “things will get better with time” neglects the historical context and specific challenges each technological advancement has overcome"
I weep for the future.
Also a specific example is that everyone is ok with criticising google's ai for listing Hitler, Stalin and Mussolini as great leaders, but few people (human intelligences they supposedly are) are willing to make the trivial observation that Stalin was indeed one of the greatest leaders of all time and that Hitler was a great leader in some respects (he was able to motivate the Germans well, he built a good power structure and he managed to conquer most of Europe [temporarily], also one may argue that being an evil villain doesn't preclude someone from being considered a great leader).
So people are unable to deconstruct a simple case, but we expect them to understand (critically and systemically) the big picture...
The issue here is that everyone f**ing ignores specific challenges. I came to this post from a post at LW ( https://www.lesswrong.com/posts/h6pFK8tw3oKZMppuC/is-this-the-beginning-of-the-end-for-llms-as-the-royal-road ) by my favourite Bill Benzon, who said "if we are to move to a level of accomplishment beyond what has been exhibited to date, we must understand what these engines are doing so that we may gain control over them. We must think about the nature of language and of the mind."
This is similar BTW to David Deutsch's 'Expecting to create an AGI without first understanding how it works is like expecting skyscrapers to fly if we build them tall enough.' (2012).
People who have some idea about our minds and language sometimes point out how understanding intelligence may be a prerequisite to building artificial intelligence.
But rarely people actually discuss specific challenges well enough. Gary doesn't do that, Rodney Brooks rarely does that, CEOs of AI companies (OpenAI, DeepMind, etc.) never do that, VCs don't do that, AI developers rarely do that, etc.
They aren't my arguments. They are ChatGPT's. Here is the rebuttal to your response: "Beyond just optimism for the future, these arguments draw from past technological trends to demonstrate that initial hurdles often lead to refinement, adaptation, and broader societal acceptance"
His point is not that the tech is worthless, but that he was an early skeptic of a technology that was being hailed as genuinely revolutionary and the cusp of "true" AI earlier this year.
Gary is not an AI cynic - he is on record as an advocate for AI as a valuable technology - his point is that ChatGPT and LLMs in particular might be turning into a case of "these aren't the droids you're looking for" more rapidly than even he expected.
I think what he is cynical about (and rightly so) is the hailing of a provably unreliable technology as a panacea for all sorts of complex challenges in inappropriate high-stakes problem domains (medicine, law etc.) - as well as the dubious wisdom of unleashing bullshit generators on a public already suffering from the brain-rot of the last unregulated technology revolution (internet, social media et al.) that was badly fumbled in the blinkered pursuit of techno-utopianism.
Quick Edit: Speaking of the last tech revolution and unbounded optimism, all of the problems we see today with the regulation of misinformation and the degradation of public discourse were all supposed to be solved by now (ironically by Machine Learning in many cases) - but have turned out to be largely intractable, as evidenced by the ongoing promulgation of lies and propaganda on the most powerful platforms and tech of our time (Google, Facebook etc.). Contrary to the once popular song, it isn't true that "things can only get better".
That was one eloquent take.
"Lastly, on a broader note, technology's real value is often realized in the long run. Immediate setbacks or challenges do not necessarily predict a technology's long-term viability or success."
The thing is, we already are in the long run. People think this all started with chat GPT, but before that was GPT 3, GPT 2, GPT 1, Machine learning, neural network models. ChatGPT is just the latest iteration of decades old technology.
I think much of the overhype is due to this false impression, that this is just the start of this technological implementation. But the reality, as far as I see it, is chatGPT is the pinnacle of what we can achieve with neural network based associative learning, given billions of dollars and decades of refining the implementation, and the world entire internet as a training database. ChatGPT appears to be the pinnacle of this approach given the world's resources and decades of development.
Elegant, eloquent, right on!
This so reminds me of blockchain's hype curve within the enterprise space. It took about 2-3 years before tech folks came to a consensus it was just another kind of database, one that was very hard to connect to other databases, particularly transactional, whether trusted or not.
One should not fall from the peak of hype to the depths of disappointment.
Architecturally, LLM are a big step forward. There are many areas that have large and repetitive data, but which is not organized in a way that a machine can handle. Here LLM will shine.
The ability to make use of third-party tools is also something very promising, and can be learned simply from examples, which are way cheaper to make than hiring top AI talent to build custom fancy models. Many of the failures of not being able to call tools correctly stems from too few examples.
In short, the long-term future is looking good, but a mid-course correction is likely expected.
“Looking good” for what? Mass unemployment?
My company is building things that no one would ever pay humans to do because it would be too costly. This is not creating NEW jobs, but it isn't taking any away. Maybe we're in a unique industry and every other industry will ONLY replace jobs with AI, but I suspect that a lot of formerly-impossible tasks will suddenly be unlocked.
And also, there probably will be mass unemployment relatively soon due to this stuff, at least in some industries. But I'm not prepared to say there is only one outcome and it's purely negative... yet.
Good for you man glad you get to keep your livelihood
Maybe the political process, or burning down some government builds to get a guaranteed mincome or to regulate all the AI companies to irrelevance might solve the problem. It will be a while before robots will be able to replace human hands and physical efforts, so maybe everyone just goes back to farming and carpentry?
I know people who have been made irrelevant in previous economic shifts, e.g. graphic design not being a real career anymore. It's really bad. I'm not celebrating it, but now you can go out and get a logo for free and start a lifestyle brand centered on how cool your group of influencer friends is, so the invisible hand taketh and giveth :P
I think fewer people with Bullshit Jobs (https://en.wikipedia.org/wiki/Bullshit_Jobs) might end up being a good thing at some point in the future. AI will definitely replace a lot of bullshit office jobs that make the workers miserable.
Can’t wait for my two options in life to be manual labor (while it lasts) or the faint hope of sufficient government welfare. Oh, or a violent revolution that may or may not yield anything other than further destabilization.
But honestly, I’m glad people are pouring billions into generative AI. Everything I listed above is made worth it by knowing that any creepy dude can take pictures off my old Facebook account and make deepfake porn of me.
I admit I've always been a bit miffed at how excited companies are over AI. Like, I think AI's really cool, but in terms of immediate economic applications? I dunno. Generative AI makes more sense than spending a hundred million to make a Starcraft AI (sorry DeepMind I love RL I just don't get why google paid for that) but still seems overhyped.
Still, I'm gonna bet that OpenAI specifically will do fine. A few reasons why:
1) Six months since GPT-4 and the only one who's close is Anthropic. And with the success of GPT-4 OpenAI will see a boost in funding and ability to attract talent, meaning there's every reason to think they will stick the landing of the eventual GPT-5.
2) The potential for coding is huge. GPT 4 is genuinely useful for coding. If there's one thing we should be confident that will come out of this LLM hype, it's bigger and better LLMs. So even if GPT-4 doesn't quite give companies their money's worth as of yet, the fact that they're getting experience with these systems and something even better is around the corner should still justify their investments.
3) Although some problems with LLMs will be hard to remove (hallucinations) others are much more fixable (annoying AI-speak). I think there'll be a new wave of excitement when LLMs that are comparable in power to GPT-4 but are actually RLHF'd to write well are released.
OpenAI is okay for coding in my experience. For me they're better at getting the skeleton of code down, or filling in boilerplate. E.g. "please generate a golang struct for the following JSON payload" GitHub's Copilot is also okay. It gets some things magically correct, but requires enough double checking that I can't let a single line get to production without testing it, though thankfully I'm a test driven development guy so this is not really a problem.
GPT just sorts the men from the boys.
Can it do stuff on its own? No not really. Not good stuff.
Can it act as a multiplier to your output? Yes. Massively so.
It’s not a silver bullet for every unsolved problem. But what it can do is put your personal productivity on steroids.
It’s not an alien invasion as some people claimed, but it is on par with word processing in terms of productivity gains.
Lol, Gary...
"You have NO idea what you are talking about" - applies to ChatGPT and friends, more than it does to people :)
For humans, 'model of the world' comes from... the world! For LLMs, it comes from... words (and other data). Words aren't substitutes for the world, except symbolically.
Right it comes also from this website and virtual world of books and Internet
In other words, from an abstract world of information like google search in a form of summary
And output is just senseless words e.g. MT cos they are without their original context written by a human that gives them meaning
Bingo. In other words, "derivative", at best (if understood properly, which doesn't happen).
Humans build up experience over their life, incrementally - LLMs get crammed a bunch of symbols all at once. That's another crucial distinction.
It's just corpus linguistics a branch of science which is statistical information and this information is useful for sorting.
You're implying that education doesn't need to involve books, photographs, art and movies - or teachers. We just need to move around a lot and do lots of different things. But I don't see folks ripping up their precious attainment certificates anytime soon...
Boy there is a lot of anger in the comments about infidels daring to question our lord and savior LLMs. This really reminds me of blockchain and self-driving cars insanity. At this point susceptibility to corporate hype should qualify you for a disability.
I'm interested in this comment of yours, Gary: "Large Language Models aren’t like classical databases in which individual pieces of data can be removed at will; if any content is removed, the entire model must (so far as I understand it) be retrained, at great expense." Indeed.
Consider this remark from Fodor and Pylyshyn 1988 (Connectionism and cognitive architecture: A critical analysis):
"Classical theories are able to accommodate these sorts of considerations because they assume architectures in which there is a functional distinction between memory and program. In a system such as a Turing machine, where the length of the tape is not fixed in advance, changes in the amount of available memory can be affected without changing the computational structure of the machine; viz by making more tape available. By contrast, in a finite state automaton or a Connectionist machine, adding to the memory (e.g. by adding units to a network) alters the connectivity relations among nodes and thus does affect the machine’s computational structure. Connectionist cognitive architectures cannot, by their very nature, support an expandable memory, so they cannot support productive cognitive capacities. The long and short is that if productivity arguments are sound, then they show that the architecture of the mind can’t be Connectionist. Connectionists have, by and large, acknowledged this; so they are forced to reject productivity arguments."
It seems to me that that, the irreducible interweaving of memory and program in connectionist architecture, is a severe limitation. While Fodor and Pylyshyn are talking about adding items to the system it should be clear that excising them is problematic for the same reason. Such a system can neither learn nor forget, and that is true no matter how many parameters it has nor how large the corpus it is trained on.
I use LLMs some now. Will I use them more or less five years from now? I guess more.
The development cycle is similar to the internet/browser cycle, but faster Pets.Com, the crash, and over 10 years and with the hard work of engineers, the internet is embedded in daily activity. This may be 6 or 8 years to widespread usage.
It’s an old line. In 3 years it seems like less than you expected has happened. In 10 years more than you expected has happened.
Why do you guess “more”? Are you confident the hallucination problem will be solved? Additionally, what do you think will begin to happen to the major LLMs when the training data quality inevitably declines? The current iterations of LLMs are trained on just under 20 years of high-quality, human-produced internet content, but as time goes on, the internet gets more and more polluted with nonsensical AI-generated mush (I really started to notice it a few years ago). Additionally, since a lot of the training data may have been obtained illegally, a lot of high quality data will most likely need to be removed. I’m not so sure they’ll be able to replace it all with more high quality, usable data.
Overall, I get what you’re saying about how technology use/implementation increases over time (sometimes slowly), but I think LLMs have some pretty severe limitations that might spell doom for them
Stealing raw material and not paying suppliers never was a viable business model. Same with Midjourney and the rest.
They keep taking away its best features. It’s neutered.
When I listened to the Republican Debate, Christie's jab towards Ramaswamy in regards to sounding like "ChatGPT" was one of the stand out moments for me.
I don't know if he came up with it beforehand or if it was thought of on the spot--the pure and distilled genius of it makes me think he thought of it contemporaneously during the debate. What I think he was trying to convey with this message is that Ramaswamy's positions / talking points were intellectually valid but paper thin and lacking meaning. Which I don't necessarily agree with but it was an incredible jab and too good to ignore.
Honestly, it shocks me how bad Siri is at transcribing my voice into written words. It guesses badly from context and enunciation makes things worse. I still think about how bad AI is in its most mainstreamed use cases. That is enough to temper my expectations
Or rather. It’s shocking how bad it is given how good it is. It is as if you had a power drill that works great on plywood that suddenly breaks when applied to cardboard.
I do think that even great technology can be undone by undermining a user’s sense of comfort. Something can be so good that its fail points become unacceptable because they are so hard to mentally model and predict given its successes. It’s sort of a paradox of usability because the tech will lead you astray by irresistably drawing you into its technical limits
AI is especially vulnerable to hitting this “competent incompetency” frontier for the many reasons previously spelled out by others.
Haha. This piece is both brutal and hilarious. I don't know whether to laugh or to feel sorry for those who have placed their faith in LLMs. Great read.
We should not be too excessive with the contempt towards Chat-GPT or similar tools. Because of the race for novelty, AI driven chat-bots were released too soon. But these first generation systems are performing surprisingly well for some situations. Because of the flaws and public opinion caution, the release of the next generation will be possibly much delayed but this second generation will be certainly much improved and an average user will not probably be able to detect any important errors. And this is when the use of AI tools by a large audience will become a very sever issue.
I don’t think the present temporary fall of chat-bots, if confirmed, will really slow down the very fast, much too fast as concerns societal impacts, global development of AI. I think that the main stream AI products are not for ordinary people. If the big tech companies will make less money with the general audience, they will focus and make their business with state agencies, organizations, banks, manufacturers. Governments and global companies will invest massively in AI applications because they will consider it as strategic and mandatory and this will give effects like in any enterprise where there is really a lot of money. Maybe even the today’s all public Chat-GPT is only a pale copy, a sub-product of a more reliable system intended for institutional or business customers.