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

Notes on OpenAI's new o1 chain-of-thought models

simonwillison.net

Post reply on HN
Print view View original

11–20 of 659 posts

Re: Notes on OpenAI's new o1 chain-of-thought models

#11

Post by echelon »

jumploops wrote: post #3

> the idea that I can run a complex prompt and have key details of how that prompt was evaluated hidden from me feels like a big step backwards. As a developer, this is highly concerning, as it makes it much harder to debug where/how the “reasoning” went wrong. The pricing is also silly, because I’m paying for tokens I can’t see. As a user, I don’t really care. LLMs are already magic boxes and I usually only care abo…

More opportunity for competitors to differentiate.

OpenAI doesn't really have a moat. This isn't payments or SMS where only Stripe or Twilio were trying to win the market. Everybody and their brother is trying to build an LLM business.

Grab some researchers, put some compute dollars in, and out comes a product.

Everyone wants this market. It's absurdly good for buyers.

Re: Notes on OpenAI's new o1 chain-of-thought models

#12

Post by OutOfHere »

ldjkfkdsjnv wrote: post #5

The theory is that this solves the data shortage problem, they can generate a ton of chain of reasoning data from what we already have. True iterative improvement, like out of a science fiction novel These models are going to get embedded deeply into IDE's, like cursor has, and essentially end software development as we know it. A properly written requirements spec, and an engineer, can do the work of 5. Software eng…

> A properly written requirements spec, and an engineer, can do the work of 5.

I do not think this will scale. GPT o1 is presumably good for bootstrapping a project using tools that the engineer is not familiar with. The model will struggle to update a sizable codebase, however, with dependencies between the files.

Secondly, no matter the size of the codebase and no matter the model used, the engineer still has to review every single line before incorporating it into the project. Only a competent engineer can review code effectively.

Re: Notes on OpenAI's new o1 chain-of-thought models

#13

Post by famouswaffles »

OutOfHere wrote: post #10
jumploops wrote: post #3

> the idea that I can run a complex prompt and have key details of how that prompt was evaluated hidden from me feels like a big step backwards. As a developer, this is highly concerning, as it makes it much harder to debug where/how the “reasoning” went wrong. The pricing is also silly, because I’m paying for tokens I can’t see. As a user, I don’t really care. LLMs are already magic boxes and I usually only care abo…

> As a user, I don’t really care. As a user, whether of ChatGPT or of the API, I absolutely do care, so I can modify and tune my prompt with the necessary clarifications. My suspicion is that the reason for hiding the reasoning tokens is to prevent other companies from creating a big CoT reasoning dataset using o1. It is anti-competitive behavior. If a user is paying through the nose for the reasoning tokens, and yes…

>My suspicion is that the reason for hiding the reasoning tokens is to prevent other companies from creating a big CoT reasoning dataset using o1.

I mean...they say as much

Re: Notes on OpenAI's new o1 chain-of-thought models

#14

Post by echelon »

famouswaffles wrote: post #13
OutOfHere wrote: post #10

Earlier quoted context omitted.

> As a user, I don’t really care. As a user, whether of ChatGPT or of the API, I absolutely do care, so I can modify and tune my prompt with the necessary clarifications. My suspicion is that the reason for hiding the reasoning tokens is to prevent other companies from creating a big CoT reasoning dataset using o1. It is anti-competitive behavior. If a user is paying through the nose for the reasoning tokens, and yes…

>My suspicion is that the reason for hiding the reasoning tokens is to prevent other companies from creating a big CoT reasoning dataset using o1. I mean...they say as much

Once again true to their name.

Re: Notes on OpenAI's new o1 chain-of-thought models

#15

Post by ldjkfkdsjnv »

OutOfHere wrote: post #12
ldjkfkdsjnv wrote: post #5

The theory is that this solves the data shortage problem, they can generate a ton of chain of reasoning data from what we already have. True iterative improvement, like out of a science fiction novel These models are going to get embedded deeply into IDE's, like cursor has, and essentially end software development as we know it. A properly written requirements spec, and an engineer, can do the work of 5. Software eng…

> A properly written requirements spec, and an engineer, can do the work of 5. I do not think this will scale. GPT o1 is presumably good for bootstrapping a project using tools that the engineer is not familiar with. The model will struggle to update a sizable codebase, however, with dependencies between the files. Secondly, no matter the size of the codebase and no matter the model used, the engineer still has to re…

I respectfully, but completely disagree. Right now with sonnet 3.5 + cursor ide, I'm not writing that much of my own code at my FAANG job. I am generating a ton, passing in documentation from internal libraries, iterating on the result. Most of the time, I just accept its changes.

This is going to rapidly happen. All we need are a few more model releases, not even a step function improvement

Re: Notes on OpenAI's new o1 chain-of-thought models

#16

Post by ldjkfkdsjnv »

bamboozled wrote: post #9
ldjkfkdsjnv wrote: post #5

The theory is that this solves the data shortage problem, they can generate a ton of chain of reasoning data from what we already have. True iterative improvement, like out of a science fiction novel These models are going to get embedded deeply into IDE's, like cursor has, and essentially end software development as we know it. A properly written requirements spec, and an engineer, can do the work of 5. Software eng…

Which will mean...there is going to be a lot more software?

Probably yeah

Re: Notes on OpenAI's new o1 chain-of-thought models

#17

Post by notarealllama »

heisenzombie wrote: post #6
GaggiX wrote: post #2

I imagine that GPT-5 would be a refined version of this paradigm, probably with omni (multimodal) capabilities added (input and output).

Fascinating, I wonder if we'll get non-textual hidden reasoning tokens? "Let me draw myself a diagram".

I know I sometimes sketch or write intermediaries before then compiling a full response.

If AI can do this on 64k tokens, iteratively, fully multimodal... I don't think I've ever actually been scared of a super intelligence / singularity moment until just now.

Now this is AI!

Re: Notes on OpenAI's new o1 chain-of-thought models

#18

Post by layer8 »

The o1-preview model still hallucinates non-existing libraries and functions for me, and is quickly wrong about facts that aren't well-represented on the web. It's the usual string of "You're absolutely correct, and I apologize for the oversight in my previous response. [Let me make another guess.]"

While the reasoning may have been improved, this doesn't solve the problem of the model having no way to assess if what it conjures up from its weights is factual or not.

Re: Notes on OpenAI's new o1 chain-of-thought models

#19

Post by OutOfHere »

ldjkfkdsjnv wrote: post #15
OutOfHere wrote: post #12

Earlier quoted context omitted.

> A properly written requirements spec, and an engineer, can do the work of 5. I do not think this will scale. GPT o1 is presumably good for bootstrapping a project using tools that the engineer is not familiar with. The model will struggle to update a sizable codebase, however, with dependencies between the files. Secondly, no matter the size of the codebase and no matter the model used, the engineer still has to re…

I respectfully, but completely disagree. Right now with sonnet 3.5 + cursor ide, I'm not writing that much of my own code at my FAANG job. I am generating a ton, passing in documentation from internal libraries, iterating on the result. Most of the time, I just accept its changes. This is going to rapidly happen. All we need are a few more model releases, not even a step function improvement

Don't you still have to explain your requirement really well to it, in a lot of detail? In a terse language like Python, I might as well just write the code. In a verbose language like Java, perhaps there is more of a value in detailing the requirement.

Re: Notes on OpenAI's new o1 chain-of-thought models

#20

Post by tptacek »

layer8 wrote: post #18

The o1-preview model still hallucinates non-existing libraries and functions for me, and is quickly wrong about facts that aren't well-represented on the web. It's the usual string of "You're absolutely correct, and I apologize for the oversight in my previous response. [Let me make another guess.]" While the reasoning may have been improved, this doesn't solve the problem of the model having no way to assess if what…

I've had the opposite experience with some coding samples. After reading Nick Carlini's post, I've gotten into the habit of powering through coding problems with GPT (where previously I'd just laugh and immediately give up) by just presenting it the errors in its code and asking it to fix them. o1 seems to be effectively screening for some of those errors (I assume it's just some, but I've noticed that the o1 things I've done haven't had obvious dumb errors like missing imports, and all my 4o attempts have).
Post reply on HN