Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

ChatGPT as a Calculator for Words

simonwillison.net

Post reply on HN
Print view View original

41–50 of 81 posts

Re: ChatGPT as a Calculator for Words

#41

Post by dragonwriter »

pmoriarty wrote: post #38
ftxbro wrote: post #2

As a long time LLM enjoyer I just want to mention https://generative.ink/posts/simulators/ as I think it's by far the most insightful take on the GPT LLMs even though it was from before ChatGPT. It's better than blurry jpeg and stochastic parrot etc.

One thing I've been impressed by recently is ChatGPT's ability to explain epigrams. How can it explain them effectively if it's just a "stochastic parrot" or a "blurry jpeg"? Neither fits what it does. Don't such effective explanations require some sort of understanding? Below is an example explanation, and there are more here: [1] --- Explain "In software systems it is often the early bird that makes the worm." This…

It literally misses any explanation of the only thing that distinguishes the epigram from the proverb it refers to. While identifying, and reproducing, with simple contextual substitutions, and explanation of the proverb referenced is not a small feat compared to where this kind of technology was a few years ago, its pretty darned far from either (1) actually explaining the epigram, or (2) making the argument you try to make based on explaining the epigram.

Re: ChatGPT as a Calculator for Words

#42

Post by tbalsam »

pretendscholar wrote: post #40
LelouBil wrote: post #12

Earlier quoted context omitted.

Can you explain more ? I know the basics of deep learning and I found the article accurate.

As an amatuer I think Markov chains are explicitly a crude frequency association whereas what exactly a neural network is storing to predict the next token involves stored representations in neural weights which can be far more nuanced.

Thank you, this is a good compressed and less-salty response. I appreciate your contribution to the conversation and will use aspects of it when trying to explain the matter in the future. <3 :thumbsup:

Re: ChatGPT as a Calculator for Words

#43

Post by fnordpiglet »

stevenhuang wrote: post #34
ftxbro wrote: post #2

As a long time LLM enjoyer I just want to mention https://generative.ink/posts/simulators/ as I think it's by far the most insightful take on the GPT LLMs even though it was from before ChatGPT. It's better than blurry jpeg and stochastic parrot etc.

Yup, that's a classic article. My favorite is Chalmer's engine bit: > What pops out of self-supervised predictive training is noticeably not a classical agent. Shortly after GPT-3’s release, David Chalmers lucidly observed that the policy’s relation to agents is like that of a “chameleon” or “engine”: >> GPT-3 does not look much like an agent. It does not seem to have goals or preferences beyond completing text, for…

Yeah that’s been my thought since the beginning. We have created all sorts of agents to date that are very compelling but lack a bridge into knowledge and synthesis of complex semantics. Wiring an ensemble of the various AI models created over the last 70 years I think is the final step.

Re: ChatGPT as a Calculator for Words

#44

Post by fomine3 »

swader999 wrote: post #28
dror wrote: post #21

Earlier quoted context omitted.

He can make the same criticism of Internet searches as he does of GPT: you shouldn't trust them until you validate them. I find that GPT's answers are for the most part more reliable the searches, specifically today's searches. In the last 12 months, search results have become so spammy with AI generated pages (oh the irony), that it's hard to find reliable answers. So like search, I look at GPT's answers with a grai…

Google and others would be wise to add a date filter of "before summer 2023". Maybe a bit longer, but not much time left till AI spam really takes over.

Spammers will set article date to 2021

Re: ChatGPT as a Calculator for Words

#45

Post by anotherpaulg »

dror wrote: post #21
SomewhatLikely wrote: post #6

The author does a good job of pointing out what may be the strongest skills of LLMs but the claim they aren't useful as a search engine didn't ring particularly true. For many questions I have ChatGPT is the best tool to use because I know the topics I'm asking about are mentioned hundreds of times in the web, and the LLM can distill down at knowledge to the specifics I'm asking about. If you treat it as a friend who…

He can make the same criticism of Internet searches as he does of GPT: you shouldn't trust them until you validate them. I find that GPT's answers are for the most part more reliable the searches, specifically today's searches. In the last 12 months, search results have become so spammy with AI generated pages (oh the irony), that it's hard to find reliable answers. So like search, I look at GPT's answers with a grai…

Gish looks really nice. I'm going to give it a try.

It seems like you've been using similar workflows to what I've been trying for coding with gpt?

https://github.com/paul-gauthier/easy-chat#created-by-chatgp...

Re: ChatGPT as a Calculator for Words

#46

Post by flangola7 »

fomine3 wrote: post #44
swader999 wrote: post #28

Earlier quoted context omitted.

Google and others would be wise to add a date filter of "before summer 2023". Maybe a bit longer, but not much time left till AI spam really takes over.

Spammers will set article date to 2021

You can't fake a domain registration date

Re: ChatGPT as a Calculator for Words

#47

Post by akhilpotla »

dror wrote: post #21
SomewhatLikely wrote: post #6

The author does a good job of pointing out what may be the strongest skills of LLMs but the claim they aren't useful as a search engine didn't ring particularly true. For many questions I have ChatGPT is the best tool to use because I know the topics I'm asking about are mentioned hundreds of times in the web, and the LLM can distill down at knowledge to the specifics I'm asking about. If you treat it as a friend who…

He can make the same criticism of Internet searches as he does of GPT: you shouldn't trust them until you validate them. I find that GPT's answers are for the most part more reliable the searches, specifically today's searches. In the last 12 months, search results have become so spammy with AI generated pages (oh the irony), that it's hard to find reliable answers. So like search, I look at GPT's answers with a grai…

Speaking of AI generated pages, I wonder how OpenAI filter these low quality web pages out of their training set as they continue to training.

Also, I wonder how they decide what code is worth training on. Because a lot of code is written in poor style/has technical debt, it might be the case that these LLMs in the long run lead to an increase in the technical debt in our society. Plus, eventually, and this might already be happening, the LLM are going to end up training on their own outputs, so that could lead to self immolation by the model. I am not certain RLHF completely resolves this issue.

Re: ChatGPT as a Calculator for Words

#48

Post by Myrmornis »

ftxbro wrote: post #2

As a long time LLM enjoyer I just want to mention https://generative.ink/posts/simulators/ as I think it's by far the most insightful take on the GPT LLMs even though it was from before ChatGPT. It's better than blurry jpeg and stochastic parrot etc.

I’d like to know who the author is and what their background/professional experience is. Is that information available? I see they use a handle “moire”.

Re: ChatGPT as a Calculator for Words

#49

Post by ftxbro »

Myrmornis wrote: post #48
ftxbro wrote: post #2

As a long time LLM enjoyer I just want to mention https://generative.ink/posts/simulators/ as I think it's by far the most insightful take on the GPT LLMs even though it was from before ChatGPT. It's better than blurry jpeg and stochastic parrot etc.

I’d like to know who the author is and what their background/professional experience is. Is that information available? I see they use a handle “moire”.

it's not me and they have wanted to be pseudonymous. pls no dox
Post reply on HN