10 ms·

Agree- so many are dismissing it because its not perfect, but this is an absolutely enormous leap from anything else I have seen in my lifetime, and this crosse

Agree- so many are dismissing it because its not perfect, but this is an absolutely enormous leap from anything else I have seen in my lifetime, and this crosses the chasm IMHO between "AI is unlikely to really be able to have general applicability" to "AI is probably going to upend life as we know it in our lifetimes." Maybe it will just peter out at this level like self driving car technology did, but I really doubt it, and it seems to me that the building blocks are already there to commercialize this and use it in a real way. The future is going to be interesting...

spaceman_2020 4y ago
ChatGPT is smart enough at the first prompt, but what really stands out - freakily, I might add - is how it can remember earlier data. You can go several levels deep and it will still reference the first answer and contextualize it properly. "Do X" is great. But "Now change X to Y in the style of Z" is what blows my mind.
- kevstev 4y ago
  Thats interesting. When I did play with it a bit, it wasn't clear to me if it was actually doing that, but in hindsight it definitely was and was not just a coincidence. When I first used ChatGPT, I had a dog that recently passed, and she had such a big personality (a mischievous always hungry hunting dog) and was just very different than my other dog (a pitbull mix that likes to guard things), and I had thought that the two of them would be a good basis for a children's book. I initially put in a prompt like "tell me a story about the hunting dog and the guard dog." and the results were pretty meh. I was interested to see if it was actually pulling in data from FB so I put her name in instead (she had an FB page yes). It did not seem to do much. So I put in some things like "the little hunting dog is always hungry." "the hunting dog is always looking to hunt" and then did a fuller prompt of "the hunting dog and the guard dog are going to the park. tell me a story about it" and finally got a very good fully fleshed out story about how the little dog was on the prowl for a squirrel while the guard dog watched out and it was really well written and made about as much sense as a fictional tale could. There was no awkwardness or other tells that would have made you think that this was written by anything other than a human. Anyway, that is huge that it can do this. I do remember when I used to play with other chatbots that remembering past context was something they were unable to do, at least not well, but in most cases it seems they didn't even attempt to. ChatGPT has blown me away in many ways, but this is a very specific leap forward.
  - spaceman_2020 4y ago
    It usually always comes up with something very meh the first time around. You have to nudge it in the right direction. I was showing it off to my wife. I asked it to write a lecture for an intro to Latin American poetry with a focus on Pablo Neruda, including a background on Chilean history. The first result was just 5-6 paragraphs with the most rudimentary 101 on all the above topics. Then I told it to rewrite it as if the lecture was going to be given to grad students. Next, I told it to include examples of Neruda's poetry along with some analysis. I also asked it to flesh out the segment on Chilean history, with a focus on the independence movement. The final lecture was good enough that you could walk into a classroom of college seniors and deliver it without any edits. The first one might have been good enough for 9th graders at most.
- 93po 4y ago
  In my experience, ChatGPT was perfectly happy providing logic for a question that counteracted previous logic it provided. When I pointed out this conflict, it disagreed that it was in conflict when it clearly was.
sumtechguy 4y ago
One of them I used was supposed to be 'pick-a-path' style fantasy fight the dragon sort of thing. I decided to throw the A-Team into it. It integrated them in and even knew the correct names and their context of what each char tended to do. By far not perfect. But the fact it could do even 'kind of' that. Is very interesting. 'kind of' is for some things is perfectly fine. The generated art ones are very interesting too. It reminds me a lot of early emulation discussions. Where people were arguing unless you had cycle perfect emulation nothing would work. Turns out 'kind of' close works for like 95% of the cases. Some things did need that sort of detail. But most didnt.
CamperBob2 4y ago
Agree- so many are dismissing it because its not perfect In fact, I'd submit that phenomenon as its own entry in the Weirdest or Most Surreal Tech Trend contest. People who, when confronted with a talking dog, say things like, "This talking dog is an idiot. It thinks 429 is prime, its painting looks like something from Bob Ross on acid, and the C++ code it wrote for me is full of bugs. I don't see what all the hype is about." Some of this dismissiveness is just whistling past the graveyard, but much of it comes from people who genuinely think the human brain is something that operates by means of ethereal, ineffable, and unreproducible mystic forces.
- kevstev 4y ago
  I feel this way about self driving cars as well. Its like even 10 years ago, the capabilities they have today would be considered mind blowing, yet they receive so much hate because they aren't yet adapted to all situations. Its baffling to me, every time I watch a video of one I get all giddy and I am like we are living in the future! Yet even from tech circles, they get so much criticism because they aren't 100% there yet. Its just a mindset I don't understand. And on places like reddit there is definitely some astroturfing going on or at least some definite "camps" that don't like SDC technology, but on HN AFAICT its been genuine disdain. So weird.
  - daveguy 4y ago
    I think most of the criticism with self driving is around the over-hyped under-delivered "full self driving" from Musk -- it actually seems to be getting worse compared to 5 years ago. Apparently Cruise and Waymo are doing much more impressive things, but they only get a fraction of the press.
    - benlivengood 4y ago
      I'm curious how many people who criticize full self drive have actually ridden in the car while it navigates city streets. It makes about as many mistakes as a 1-3 years of practice new driver. I don't own a Tesla so I've ridden with friends. It's a barely noticeable difference between human and FSD operation, with very few overrides necessary.
ozim 4y ago
Many are dismissing it because it is just neural network. Which were basically at academic works since 80’s or earlier. It is only that now we have enough hardware and material for training neural nets that they start to be useful.
- btilly 4y ago
  There is a lot more than that. 1. We've learned that relu is a really good function to use for neural networks. 2. We've learned that regularization is essential to having large networks without overfitting. 3. We've learned how to scale models up and down so that you can train something on a giant cluster then run it on your phone. 4. We've learned how to apply this to lots of specific kinds of problems. etc, etc, etc. We do NOT simply do what we did in the 80s then throw more hardware at it. Instead we are getting better at making the hardware do something useful.
ActorNightly 4y ago
>nd this crosses the chasm IMHO between "AI is unlikely to really be able to have general applicability" to "AI is probably going to upend life as we know it in our lifetimes." ChatGPT3 isn't really the definitive sign of the latter. Its just a stepping stone on the way. As soon as humans discovered differential calculus, that was the genesis to the eventual development of a general AI. Everything else along the way is an issue of optimization and spending raw compute time, whether its done by humans or machines.
jredwards 4y ago
I had the same reaction. It's the first time in my life where I've ever thought, "This is the beginning of an era. They'll give a name to this period of time." And yeah, maybe that's overselling it, but I had that thought. I've seen so many pieces of incredible technology struggle to find a use-case. By contrast, ChatGPT almost has too many use-cases to catalog, and while it's far from perfect, it seems obviously, to me, to be a tipping point of sorts.

spaceman_2020 4y ago

ChatGPT is smart enough at the first prompt, but what really stands out - freakily, I might add - is how it can remember earlier data. You can go several levels deep and it will still reference the first answer and contextualize it properly. "Do X" is great. But "Now change X to Y in the style of Z" is what blows my mind.

kevstev 4y ago
Thats interesting. When I did play with it a bit, it wasn't clear to me if it was actually doing that, but in hindsight it definitely was and was not just a coincidence. When I first used ChatGPT, I had a dog that recently passed, and she had such a big personality (a mischievous always hungry hunting dog) and was just very different than my other dog (a pitbull mix that likes to guard things), and I had thought that the two of them would be a good basis for a children's book. I initially put in a prompt like "tell me a story about the hunting dog and the guard dog." and the results were pretty meh. I was interested to see if it was actually pulling in data from FB so I put her name in instead (she had an FB page yes). It did not seem to do much. So I put in some things like "the little hunting dog is always hungry." "the hunting dog is always looking to hunt" and then did a fuller prompt of "the hunting dog and the guard dog are going to the park. tell me a story about it" and finally got a very good fully fleshed out story about how the little dog was on the prowl for a squirrel while the guard dog watched out and it was really well written and made about as much sense as a fictional tale could. There was no awkwardness or other tells that would have made you think that this was written by anything other than a human. Anyway, that is huge that it can do this. I do remember when I used to play with other chatbots that remembering past context was something they were unable to do, at least not well, but in most cases it seems they didn't even attempt to. ChatGPT has blown me away in many ways, but this is a very specific leap forward.
- spaceman_2020 4y ago
  It usually always comes up with something very meh the first time around. You have to nudge it in the right direction. I was showing it off to my wife. I asked it to write a lecture for an intro to Latin American poetry with a focus on Pablo Neruda, including a background on Chilean history. The first result was just 5-6 paragraphs with the most rudimentary 101 on all the above topics. Then I told it to rewrite it as if the lecture was going to be given to grad students. Next, I told it to include examples of Neruda's poetry along with some analysis. I also asked it to flesh out the segment on Chilean history, with a focus on the independence movement. The final lecture was good enough that you could walk into a classroom of college seniors and deliver it without any edits. The first one might have been good enough for 9th graders at most.
93po 4y ago
In my experience, ChatGPT was perfectly happy providing logic for a question that counteracted previous logic it provided. When I pointed out this conflict, it disagreed that it was in conflict when it clearly was.

sumtechguy 4y ago

One of them I used was supposed to be 'pick-a-path' style fantasy fight the dragon sort of thing. I decided to throw the A-Team into it. It integrated them in and even knew the correct names and their context of what each char tended to do. By far not perfect. But the fact it could do even 'kind of' that. Is very interesting. 'kind of' is for some things is perfectly fine. The generated art ones are very interesting too. It reminds me a lot of early emulation discussions. Where people were arguing unless you had cycle perfect emulation nothing would work. Turns out 'kind of' close works for like 95% of the cases. Some things did need that sort of detail. But most didnt.

CamperBob2 4y ago

Agree- so many are dismissing it because its not perfect In fact, I'd submit that phenomenon as its own entry in the Weirdest or Most Surreal Tech Trend contest. People who, when confronted with a talking dog, say things like, "This talking dog is an idiot. It thinks 429 is prime, its painting looks like something from Bob Ross on acid, and the C++ code it wrote for me is full of bugs. I don't see what all the hype is about." Some of this dismissiveness is just whistling past the graveyard, but much of it comes from people who genuinely think the human brain is something that operates by means of ethereal, ineffable, and unreproducible mystic forces.

kevstev 4y ago
I feel this way about self driving cars as well. Its like even 10 years ago, the capabilities they have today would be considered mind blowing, yet they receive so much hate because they aren't yet adapted to all situations. Its baffling to me, every time I watch a video of one I get all giddy and I am like we are living in the future! Yet even from tech circles, they get so much criticism because they aren't 100% there yet. Its just a mindset I don't understand. And on places like reddit there is definitely some astroturfing going on or at least some definite "camps" that don't like SDC technology, but on HN AFAICT its been genuine disdain. So weird.
- daveguy 4y ago
  I think most of the criticism with self driving is around the over-hyped under-delivered "full self driving" from Musk -- it actually seems to be getting worse compared to 5 years ago. Apparently Cruise and Waymo are doing much more impressive things, but they only get a fraction of the press.
  - benlivengood 4y ago
    I'm curious how many people who criticize full self drive have actually ridden in the car while it navigates city streets. It makes about as many mistakes as a 1-3 years of practice new driver. I don't own a Tesla so I've ridden with friends. It's a barely noticeable difference between human and FSD operation, with very few overrides necessary.

ozim 4y ago

Many are dismissing it because it is just neural network. Which were basically at academic works since 80’s or earlier. It is only that now we have enough hardware and material for training neural nets that they start to be useful.

btilly 4y ago
There is a lot more than that. 1. We've learned that relu is a really good function to use for neural networks. 2. We've learned that regularization is essential to having large networks without overfitting. 3. We've learned how to scale models up and down so that you can train something on a giant cluster then run it on your phone. 4. We've learned how to apply this to lots of specific kinds of problems. etc, etc, etc. We do NOT simply do what we did in the 80s then throw more hardware at it. Instead we are getting better at making the hardware do something useful.

ActorNightly 4y ago

>nd this crosses the chasm IMHO between "AI is unlikely to really be able to have general applicability" to "AI is probably going to upend life as we know it in our lifetimes." ChatGPT3 isn't really the definitive sign of the latter. Its just a stepping stone on the way. As soon as humans discovered differential calculus, that was the genesis to the eventual development of a general AI. Everything else along the way is an issue of optimization and spending raw compute time, whether its done by humans or machines.

jredwards 4y ago

I had the same reaction. It's the first time in my life where I've ever thought, "This is the beginning of an era. They'll give a name to this period of time." And yeah, maybe that's overselling it, but I had that thought. I've seen so many pieces of incredible technology struggle to find a use-case. By contrast, ChatGPT almost has too many use-cases to catalog, and while it's far from perfect, it seems obviously, to me, to be a tipping point of sorts.