Livestream developer preview link in case you missed it - https://www.youtube.com/live/outcGtbnMuQ
GPT-4
471–480 of 1001 posts
Re: GPT-4
#472A class of problem that GPT-4 appears to still really struggle with is variants of common puzzles. For example: >Suppose I have a cabbage, a goat and a lion, and I need to get them across a river. I have a boat that can only carry myself and a single other item. I am not allowed to leave the cabbage and lion alone together, and I am not allowed to leave the lion and goat alone together. How can I safely get all three…
It's a good observation. Although on the flip side, I almost went to type up a reply to you explaining why you were wrong and why bringing the goat first is the right solution. Until I realized I misread what your test was when I skimmed your comment. Likely the same type of mistake GPT-4 made when "seeing" it. Intuitively, I think the answer is that we do have two types of thinking. The pattern matching fast thinkin…
Systenatic analytical thinking is just the first type applied in a loop with some extra prompt rules.
Re: GPT-4
#473Genuinely surprised by the positive reaction about how exciting this all is. You ever had to phone a large business to try and sort something out, like maybe a banking error, and been stuck going through some nonsense voice recognition menu tree that doesn't work? Well imagine chat GPT with a real time voice and maybe a fake, photorealistic 3D avatar and having to speak to that anytime you want to speak to a doctor,…
People here aren’t thinking about what other people’s chatbots will do to them. They’re thinking about what chatbots they themselves can unleash upon the world.
Re: GPT-4
#474A class of problem that GPT-4 appears to still really struggle with is variants of common puzzles. For example: >Suppose I have a cabbage, a goat and a lion, and I need to get them across a river. I have a boat that can only carry myself and a single other item. I am not allowed to leave the cabbage and lion alone together, and I am not allowed to leave the lion and goat alone together. How can I safely get all three…
I think you may have misstated the puzzle. It's ok to leave the lion and the cabbage together, assuming it's not a vegetarian lion.
Which to be fair is what most people reading that problem understood the first time. I wonder what would happen if you then tell gpt "No, it's the lion that can't be with the cabbage, not the goat, try again"
Re: GPT-4
#475Genuinely surprised by the positive reaction about how exciting this all is. You ever had to phone a large business to try and sort something out, like maybe a banking error, and been stuck going through some nonsense voice recognition menu tree that doesn't work? Well imagine chat GPT with a real time voice and maybe a fake, photorealistic 3D avatar and having to speak to that anytime you want to speak to a doctor,…
I agree. This tech is awesome and has countless great uses, but I think people are really underestimating how much it is going to be used to make our collective lives worse because using it will make someone a few extra dollars.
Re: GPT-4
#476Genuinely surprised by the positive reaction about how exciting this all is. You ever had to phone a large business to try and sort something out, like maybe a banking error, and been stuck going through some nonsense voice recognition menu tree that doesn't work? Well imagine chat GPT with a real time voice and maybe a fake, photorealistic 3D avatar and having to speak to that anytime you want to speak to a doctor,…
Yea, I read all about it in Anathem over a decade ago. I've come to terms with it. We'll have a balkanized "net", the old internet will be fun garbage. Hopefully it'll cut down on the damage anonymous users and individual state interests can do. Hopefully it'll help take away the free megaphones from idiots and evil doers.
Re: GPT-4
#477Genuinely surprised by the positive reaction about how exciting this all is. You ever had to phone a large business to try and sort something out, like maybe a banking error, and been stuck going through some nonsense voice recognition menu tree that doesn't work? Well imagine chat GPT with a real time voice and maybe a fake, photorealistic 3D avatar and having to speak to that anytime you want to speak to a doctor,…
I agree. My gut reaction to previous GPT releases was interest, but for this one (before even reading it) it was dread.
I think we're very close to an inflection point where functionally all information is polluted by the possibility that it's completely hallucinated or built on something hallucinated. We're already getting there in some ways - google vs. seo, astroturfed forums, fabricated publications, and this is just that but way worse. Probably orders of magnitude worse in terms of exposed information surface.
It's basically a pollution - and one that's nearly impossible to clean. The ecosystem of referential information now has its version of microplastics.
Re: GPT-4
#478A class of problem that GPT-4 appears to still really struggle with is variants of common puzzles. For example: >Suppose I have a cabbage, a goat and a lion, and I need to get them across a river. I have a boat that can only carry myself and a single other item. I am not allowed to leave the cabbage and lion alone together, and I am not allowed to leave the lion and goat alone together. How can I safely get all three…
It's a good observation. Although on the flip side, I almost went to type up a reply to you explaining why you were wrong and why bringing the goat first is the right solution. Until I realized I misread what your test was when I skimmed your comment. Likely the same type of mistake GPT-4 made when "seeing" it. Intuitively, I think the answer is that we do have two types of thinking. The pattern matching fast thinkin…
Isn't GPT essentially tool for rephrasing what it finds on the Internet, it doesn't really think?
Re: GPT-4
#479What I don't understand is how GPT-4 is able to do reasonably well on tests like the AMC12: Many of the AMC12 questions require a number of logical/deductive steps. If GPT-4 is simply trained on a large corpus of text, how is it able to do this? Does this imply that there is some emergent deductive ability that you get simply by learning "language?" Or am I missing something? Obviously, I'm assuming that GPT-4 wasn't…
They do leverage emergent abstractions. For example, in [1] a transformer model learns the coset structure of a group to better grok its multiplication table.
[1] https://mathai-iclr.github.io/papers/papers/MATHAI_29_paper....
Re: GPT-4
#480Live demo happening now! https://www.youtube.com/live/outcGtbnMuQ