5 ms·

I kind of feel like poking at the failings if ChatGPT misses the point a bit. Yes it's certainly not an AGI or even super close but to even converse with human

I kind of feel like poking at the failings if ChatGPT misses the point a bit. Yes it's certainly not an AGI or even super close but to even converse with humans at this level is mind boggling. 10 years before stable diffusion AI could just about label pictures, now it can do, well, stable diffusion. The pace or progress is insane. Like this, I feel we might engage in a naysaying dialogue with consecutive generations of GPT-like models, but finding increasingly minor nitpicks. "Ah but does it understand diminutives"? "It's handling of sarcasm isn't up to scratch". "I tried 10 languages to converse in and Esperanto was quite weak". And then one day we might wake up to a world where we can't really nitpick anymore.

mnd999 4y ago
> The pace or progress is insane. The bullshit machine got more convincing. I guess that’s a form of progress.
- ogogmad 4y ago
  I once asked an earlier version of GPT a question that it was never asked before, and it will never be asked again, and it gave multiple imaginative and plausible answers to it. It's not a bullshit machine.
  - mnd999 4y ago
    Giving imaginative and plausible answers to something you know nothing about is the definition of bullshitting.
jvm___ 4y ago
Me and my kids nitpick, but then we say "ya, but a DOG made this" We used to say that about Dalle, now it's about ChatGPT.
joe_the_user 4y ago
No doubt the pace of progress has been remarkable. But I feel like arguments that cite only this progress make the tacit assumption that there's a single intelligence level that's progressing. That is, because large language models are getting better, they must be getting better in all imaginable skills and ability. Because their strengths are getting stronger, automatically they will overcome their weaknesses. As a counterpoint, I'd mention the failure (so-far) of self-driving cars. These constructs were impressive ten years and in various measures I'm sure have only gotten more impressive yet they still don't have a level of reliability that would allow them on the road. And in my playing ChatGTP, it is certainly quite impressive yet also puts out some nonsense with nearly every paragraph in answers to questions, including things in no way "trick questions" (Edit: one could argue that the nitpicks do mask this problem, since one doesn't need trick problems to see it). Mind-you, I'm not saying these systems can't overcome their weaknesses, I'm saying that linear progress by itself doesn't imply they'll overcome their weaknesses. Edit: I've clarified the text as I've gone.
- rich_sasha 4y ago
  The self driving car is a great example, you're right, it was so good and yet never lived up to the hype. Perhaps one difference is that a human could potentially get extremely good at textual tasks with nothing but text to learn from. You can read how to solve cryptic crosswords, see examples and extrapolate. In that sense language models have a somewhat complete training dataset. Yes this requires an understanding of the material, rather than just parroting, but the signal is there if you can separate it from noise. Driving a car requires an understanding of a much wider context which is perhaps hard to acquire with just driving data. Understanding of rain, birds on the road, shaky drivers, balls rolling out from between cars, lane restrictions... You can't just throw petabytes of data at the problem. Training data is limited and expensive, and I believe we are mostly tackling AI-assired driving with rule-based approaches. I believe self driving works just fine in simulations where data is effectively unlimited. But then it doesn't generalise to the real world where context matters.
  - joe_the_user 4y ago
    Well, I would say language isn't a single task but a system, process or tool that's flexible enough to aid in many tasks at many different levels. It be used to signal social status and education or to guide someone through fixing a flat tired and often its used on these multiple level simultaneously. One can be succeed at one level of using language without succeed at another level. But we humans expect another human to succeed or to fail fairly uniformly - or we call them "a bullshit artist". These expectations may not be met with large language models. In that sense language models have a somewhat complete training dataset. Yes this requires an understanding of the material, rather than just parroting, but the signal is there if you can separate it from noise. --> I'd agree that there's conceivably more that can be done with a language dataset. But the training process of transformers-based models isn't really oriented to engaging in the process you describe. It treated language as data and it's fundamentally a very sophisticated database that only appears to engage in such logic. As it will tell you.
    - rich_sasha 4y ago
guax 4y ago
People are nitpicking as a response from a vocal group that loves to spew the doom of every knowledge worker out there. As if this will replace programmers, doctors, writers, copywriters, etc. Looking for reasons why it would not its only natural, For that purpose, I think it still has fundamental flaws that are not solvable as easily as some seem to believe. People are also impressed, given how much is being used. Being impressed, I prefer now to know and explore its boundaries. Is this really going to a place where it will replace those workers or are those limitations a fundamental barrier to what it can do based on the method in which it works? Its strengths on my tests so far: * Summary of content for specific questions * Language learning reference and translation * Rephrasing and correction of grammar in text (paragraphs at most) Its weaknesses: * Trust of results in complex responses. (clear wrong answers) * Give references. * Ambiguous questions and clarifications. (nitpick, I think its fine as it is) * New ideas or anything thats not been documented and done before or instructed in prompt (duh) This last weakness is the crux of what annoys people so much, its a predictive language model, not AGI. I don't think it's anywhere near close to replacing any worker, supporters (I am one) should focus on what this really can do, which is to increase productivity and being an incredible tool. ps. I asked it to rewrite this response and it tends to prefer the passive voice as if its writing and article. After a few tries it didn't give me a good result I could just replace what I wrote here. It doesn't really understand what I wrote, it just rephrases in its preferred form (article type constructions). Its still super helpful to "unblock" a hard to write paragraph for me, a non native speaker of english.
- PeterisP 4y ago
  Increasing productivity == replacing workers. If a tool enables a team of 4 do the same things that previously was done by a team of 5, the tool replaced a worker.
  - notahacker 4y ago
    > If a tool enables a team of 4 do the same things that previously was done by a team of 5, the tool replaced a worker Only if you believe that demand for work to be done is fixed as the cost of doing it goes down, a belief that was last reasonable to hold about three centuries ago...
inopinatus 4y ago
My benchmark for AGI is when the machines play one another at an incomprehensible game of their own devising.
iforgotpassword 4y ago
Sorry what? With all this excitement, hype and overconfidence because we made notable progress, it is extremely important we highlight the shortcomings of "AI" by finding striking, easy to grasp examples. When co-pilot and now chatgpt showed up and managed to produce working code snippets for simple text prompts, every manager on this planet with a background in economics probably started having wet dreams about replacing every programmer in their company by AI and getting a golden name plate for their desk with all the money saved. Explaining how there is a risk that the generated code might contain logic bugs, memory safety bugs, is way too abstract for these kinds of people and then tempting to ignore, so you need to demonstrate fail modes in an accessible way. TFA did exactly that.
dandelany 4y ago
I think the pace of progress is a direct result of continually poking at the failings of each successive model.
patates 4y ago
> 10 years before stable diffusion(,) AI could just about label pictures Is there an AI that can accurately label pictures? The examples I've seen so far can recognize a few animals and objects and that's it. There are people in that photo, sure, but are they wearing a jacket or a bikini? Are they happy or sad? Which model and make is that car? Google Photos seems to do a better job at this in general but doesn't expose the tags, you can only search them.
- practice9 4y ago
  CLIP, BLIP, or a combination of them? https://huggingface.co/spaces/pharma/CLIP-Interrogator https://huggingface.co/spaces/pharma/CLIP-Interrogator But it doesn't describe every object in the image
jamespwilliams 4y ago
I agree, and say roughly this in the article: > I guess cryptic crosswords fall into the (surprisingly small) category of things that ChatGPT just isn’t very good at!
- rich_sasha 4y ago
  It's a great (and very entertaining!) article, I was more responding to the general commenter than arguing with your content.
otabdeveloper4 4y ago
> The pace or progress is insane. Behavior of "AI" isn't getting more complex, just closer to our preconceptions of "human-like" behavior. This is a key point, because this isn't the progress you wanted or asked for.
LoveMortuus 4y ago
I like how for the three examples that you gave, I don't know what diminutives are, quite often I miss sarcasm and fail to handle it and I can only speak two languages fluently and three languages very weakly. I'd say that, while for the HN crowd this isn't an AGI, for the majority of population it not only is Artificial General Intelligence, for many it's much smarter then them. The only real give away is poor handling of unwanted or misunderstood queries. Not to mention the correctness of grammar, while for my language (Slovenian) it still fails at some obvious points, the sentences and structure is already much better then most message that I've received from highschool and university students.
janalsncm 4y ago
Yes, the AI effect is real. As soon as computers can do a thing it’s no longer “AI”. But I don’t think this is a nitpick at all. GPT models hallucinate information. They are right surprisingly often, but they’re also wrong quite often too. And the problem is they are just as confident in either case. This is a fundamental, irreconcilable issue with statistical language models. They have no grounding in auditable facts. They can memorize and generate in very plausible ways but they don’t seem to have a concrete model of the world. Ask ChatGPT to play chess. It can generate a text based board and prompt you for moves, but it can’t reliably update its board correctly or even find legal moves. Note that I don’t expect it to play good moves, but the fact that it can’t even play legal moves should tell us something about its internal state. Now that GPT3 has trained on the whole internet, we may have reached a practical limit to how far you can get by simply training on more data with 1 or 2 orders of magnitude more parameters. There’s only so far you can get by memorizing the textbook. At a more practical level, for most professions “pretty good” isn’t good enough. It’s not good enough to have code that’s right 90% of the time but broken (or worse, has subtle bugs) the rest of the time.
- Aeolos 4y ago
  Humans hallucinate information and often get things wrong in ways that have no grounding in auditable facts either. The fact that a textbox can do so many diverse tasks _well_, should give everyone pause. Here's a few things it was able to do when I tested it: - generate working code in multiple programming languages (C++, Rust, Typescript, Python) - rewrite terraform tf to equivalent kubernetes yaml - accurately describe esoteric knowledge related to medical imaging - find and suggest improvements in code written by senior programmers - rewrite and improve the copy of a website - create a decent presentation outline for a VC investor pitch - suggest valid improvements to sample startup mission and vision statements - expand bullet points into a proper email that I could send out to third parties without any questions raised How many people are there in the world that can do all or even some of the above at a decent level of expertise?
  - Jensson 4y ago
    > How many people are there in the world that can do all or even some of the above at a decent level of expertise? If you tell them what to do, then correct them about all the things they are wrong about, then a lot of people can do all of those as long as they have access to Google. And then once those people have done that a while they will be able to continue doing those things without your feedback. But ChatGPT can't. This makes it fundamentally different from any human.

mnd999 4y ago

> The pace or progress is insane. The bullshit machine got more convincing. I guess that’s a form of progress.

ogogmad 4y ago
I once asked an earlier version of GPT a question that it was never asked before, and it will never be asked again, and it gave multiple imaginative and plausible answers to it. It's not a bullshit machine.
- mnd999 4y ago
  Giving imaginative and plausible answers to something you know nothing about is the definition of bullshitting.

jvm___ 4y ago

Me and my kids nitpick, but then we say "ya, but a DOG made this" We used to say that about Dalle, now it's about ChatGPT.

joe_the_user 4y ago

No doubt the pace of progress has been remarkable. But I feel like arguments that cite only this progress make the tacit assumption that there's a single intelligence level that's progressing. That is, because large language models are getting better, they must be getting better in all imaginable skills and ability. Because their strengths are getting stronger, automatically they will overcome their weaknesses. As a counterpoint, I'd mention the failure (so-far) of self-driving cars. These constructs were impressive ten years and in various measures I'm sure have only gotten more impressive yet they still don't have a level of reliability that would allow them on the road. And in my playing ChatGTP, it is certainly quite impressive yet also puts out some nonsense with nearly every paragraph in answers to questions, including things in no way "trick questions" (Edit: one could argue that the nitpicks do mask this problem, since one doesn't need trick problems to see it). Mind-you, I'm not saying these systems can't overcome their weaknesses, I'm saying that linear progress by itself doesn't imply they'll overcome their weaknesses. Edit: I've clarified the text as I've gone.

rich_sasha 4y ago
The self driving car is a great example, you're right, it was so good and yet never lived up to the hype. Perhaps one difference is that a human could potentially get extremely good at textual tasks with nothing but text to learn from. You can read how to solve cryptic crosswords, see examples and extrapolate. In that sense language models have a somewhat complete training dataset. Yes this requires an understanding of the material, rather than just parroting, but the signal is there if you can separate it from noise. Driving a car requires an understanding of a much wider context which is perhaps hard to acquire with just driving data. Understanding of rain, birds on the road, shaky drivers, balls rolling out from between cars, lane restrictions... You can't just throw petabytes of data at the problem. Training data is limited and expensive, and I believe we are mostly tackling AI-assired driving with rule-based approaches. I believe self driving works just fine in simulations where data is effectively unlimited. But then it doesn't generalise to the real world where context matters.
- joe_the_user 4y ago
  Well, I would say language isn't a single task but a system, process or tool that's flexible enough to aid in many tasks at many different levels. It be used to signal social status and education or to guide someone through fixing a flat tired and often its used on these multiple level simultaneously. One can be succeed at one level of using language without succeed at another level. But we humans expect another human to succeed or to fail fairly uniformly - or we call them "a bullshit artist". These expectations may not be met with large language models. In that sense language models have a somewhat complete training dataset. Yes this requires an understanding of the material, rather than just parroting, but the signal is there if you can separate it from noise. --> I'd agree that there's conceivably more that can be done with a language dataset. But the training process of transformers-based models isn't really oriented to engaging in the process you describe. It treated language as data and it's fundamentally a very sophisticated database that only appears to engage in such logic. As it will tell you.
  - rich_sasha 4y ago

guax 4y ago

People are nitpicking as a response from a vocal group that loves to spew the doom of every knowledge worker out there. As if this will replace programmers, doctors, writers, copywriters, etc. Looking for reasons why it would not its only natural, For that purpose, I think it still has fundamental flaws that are not solvable as easily as some seem to believe. People are also impressed, given how much is being used. Being impressed, I prefer now to know and explore its boundaries. Is this really going to a place where it will replace those workers or are those limitations a fundamental barrier to what it can do based on the method in which it works? Its strengths on my tests so far: * Summary of content for specific questions * Language learning reference and translation * Rephrasing and correction of grammar in text (paragraphs at most) Its weaknesses: * Trust of results in complex responses. (clear wrong answers) * Give references. * Ambiguous questions and clarifications. (nitpick, I think its fine as it is) * New ideas or anything thats not been documented and done before or instructed in prompt (duh) This last weakness is the crux of what annoys people so much, its a predictive language model, not AGI. I don't think it's anywhere near close to replacing any worker, supporters (I am one) should focus on what this really can do, which is to increase productivity and being an incredible tool. ps. I asked it to rewrite this response and it tends to prefer the passive voice as if its writing and article. After a few tries it didn't give me a good result I could just replace what I wrote here. It doesn't really understand what I wrote, it just rephrases in its preferred form (article type constructions). Its still super helpful to "unblock" a hard to write paragraph for me, a non native speaker of english.

PeterisP 4y ago
Increasing productivity == replacing workers. If a tool enables a team of 4 do the same things that previously was done by a team of 5, the tool replaced a worker.
- notahacker 4y ago
  > If a tool enables a team of 4 do the same things that previously was done by a team of 5, the tool replaced a worker Only if you believe that demand for work to be done is fixed as the cost of doing it goes down, a belief that was last reasonable to hold about three centuries ago...

inopinatus 4y ago

My benchmark for AGI is when the machines play one another at an incomprehensible game of their own devising.

iforgotpassword 4y ago

Sorry what? With all this excitement, hype and overconfidence because we made notable progress, it is extremely important we highlight the shortcomings of "AI" by finding striking, easy to grasp examples. When co-pilot and now chatgpt showed up and managed to produce working code snippets for simple text prompts, every manager on this planet with a background in economics probably started having wet dreams about replacing every programmer in their company by AI and getting a golden name plate for their desk with all the money saved. Explaining how there is a risk that the generated code might contain logic bugs, memory safety bugs, is way too abstract for these kinds of people and then tempting to ignore, so you need to demonstrate fail modes in an accessible way. TFA did exactly that.

dandelany 4y ago

I think the pace of progress is a direct result of continually poking at the failings of each successive model.

patates 4y ago

> 10 years before stable diffusion(,) AI could just about label pictures Is there an AI that can accurately label pictures? The examples I've seen so far can recognize a few animals and objects and that's it. There are people in that photo, sure, but are they wearing a jacket or a bikini? Are they happy or sad? Which model and make is that car? Google Photos seems to do a better job at this in general but doesn't expose the tags, you can only search them.

practice9 4y ago
CLIP, BLIP, or a combination of them? https://huggingface.co/spaces/pharma/CLIP-Interrogator https://huggingface.co/spaces/pharma/CLIP-Interrogator But it doesn't describe every object in the image

jamespwilliams 4y ago

I agree, and say roughly this in the article: > I guess cryptic crosswords fall into the (surprisingly small) category of things that ChatGPT just isn’t very good at!

rich_sasha 4y ago
It's a great (and very entertaining!) article, I was more responding to the general commenter than arguing with your content.

otabdeveloper4 4y ago

> The pace or progress is insane. Behavior of "AI" isn't getting more complex, just closer to our preconceptions of "human-like" behavior. This is a key point, because this isn't the progress you wanted or asked for.

LoveMortuus 4y ago

I like how for the three examples that you gave, I don't know what diminutives are, quite often I miss sarcasm and fail to handle it and I can only speak two languages fluently and three languages very weakly. I'd say that, while for the HN crowd this isn't an AGI, for the majority of population it not only is Artificial General Intelligence, for many it's much smarter then them. The only real give away is poor handling of unwanted or misunderstood queries. Not to mention the correctness of grammar, while for my language (Slovenian) it still fails at some obvious points, the sentences and structure is already much better then most message that I've received from highschool and university students.

janalsncm 4y ago

Yes, the AI effect is real. As soon as computers can do a thing it’s no longer “AI”. But I don’t think this is a nitpick at all. GPT models hallucinate information. They are right surprisingly often, but they’re also wrong quite often too. And the problem is they are just as confident in either case. This is a fundamental, irreconcilable issue with statistical language models. They have no grounding in auditable facts. They can memorize and generate in very plausible ways but they don’t seem to have a concrete model of the world. Ask ChatGPT to play chess. It can generate a text based board and prompt you for moves, but it can’t reliably update its board correctly or even find legal moves. Note that I don’t expect it to play good moves, but the fact that it can’t even play legal moves should tell us something about its internal state. Now that GPT3 has trained on the whole internet, we may have reached a practical limit to how far you can get by simply training on more data with 1 or 2 orders of magnitude more parameters. There’s only so far you can get by memorizing the textbook. At a more practical level, for most professions “pretty good” isn’t good enough. It’s not good enough to have code that’s right 90% of the time but broken (or worse, has subtle bugs) the rest of the time.

Aeolos 4y ago
Humans hallucinate information and often get things wrong in ways that have no grounding in auditable facts either. The fact that a textbox can do so many diverse tasks _well_, should give everyone pause. Here's a few things it was able to do when I tested it: - generate working code in multiple programming languages (C++, Rust, Typescript, Python) - rewrite terraform tf to equivalent kubernetes yaml - accurately describe esoteric knowledge related to medical imaging - find and suggest improvements in code written by senior programmers - rewrite and improve the copy of a website - create a decent presentation outline for a VC investor pitch - suggest valid improvements to sample startup mission and vision statements - expand bullet points into a proper email that I could send out to third parties without any questions raised How many people are there in the world that can do all or even some of the above at a decent level of expertise?
- Jensson 4y ago
  > How many people are there in the world that can do all or even some of the above at a decent level of expertise? If you tell them what to do, then correct them about all the things they are wrong about, then a lot of people can do all of those as long as they have access to Google. And then once those people have done that a while they will be able to continue doing those things without your feedback. But ChatGPT can't. This makes it fundamentally different from any human.