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

ChatGPT agent: bridging research and action

openai.com

Post reply on HN
Print view View original

331–340 of 508 posts

Re: ChatGPT agent: bridging research and action

#331

Post by lairv »

chaos_emergent wrote: post #302
lairv wrote: post #213

Earlier quoted context omitted.

In general most of the previous AI "breakthrough" in the last decade were backed by proper scientific research and ideas: - AlphaGo/AlphaZero (MCTS) - OpenAI Five (PPO) - GPT 1/2/3 (Transformers) - Dall-e 1/2, Stable Diffusion (CLIP, Diffusion) - ChatGPT (RLHF) - SORA (Diffusion Transformers) "Agents" is a marketing term and isn't backed by anything. There is little data available, so it's hard to have generally capa…

I disagree that there isn't an innovation. The technology for reasoning models is the ability to do RL on verifiable tasks, with the some (as-of-yet unpublished, but well-known) search over reasoning chains, with a (presumably neural) reasoning fragment proposal machine, and a (presumably neural) scoring machine for those reasoning fragments. The technology for agents is effectively the same, with some currently-in-R…

Fair enough I guess, even though the concept of agent/agentic task popped before reasoning models were really a thing

Re: ChatGPT agent: bridging research and action

#332

Post by __jonas »

WXLCKNO wrote: post #276
DanHulton wrote: post #208

Earlier quoted context omitted.

There is almost guaranteed going to be an attack along the lines of prompt-injecting a calendar invite. Those things are millions of lines long already, with tones of auto-generated text that nobody reads. Embed your injection in the middle of boring text describing the meeting prerequisites and it's as good as written in a transparent font. Then enjoy exfiltrating your victim's entire calendar and who knows what els…

In the system I'm building the main agent doesn't have access to tools and must call scoped down subagents who have one or two tools at most and always in the same category (so no mixed fetch and calendar tools). They must also return structured data to the main agent. I think that kind of isolation is necessary even though it's a bit more costly. However since the subagents have simple tasks I can use super cheap mo…

What isolation is there? If a compromised sub agent returns data that gets inserted into the main agents context (structured or not) then the end result is the same as if the main agent was directly interacting with the compromising resource is it not?

Re: ChatGPT agent: bridging research and action

#333

Post by satyrun »

kingnothing wrote: post #123
mclau157 wrote: post #84

Earlier quoted context omitted.

the bigger takeaway here is will his boss allow him to walk his dog or will he see available downtime and try to fill it with more work?

95% of people doing his job will lose them. 1 person will figure out the 2% that requires a human in the loop.

I do this kind of job and there is no way I am doing this job in 5-10 years.

I don't even think it is my company that is going to adapt to let me go but it is going to be an AI first competitor that puts the company I work for out of business completely.

There are all these massively inefficient dinosaur companies in the economy that are running digitized versions of paper shuffling and a huge number of white collar bullshit jobs built on top of digitized paper shuffling.

Wage inflation has been eating away at the bottom line on all these businesses since Covid and we are going to have a dinosaur company mass extinction event in the next recession.

IMO the category error being made is that LLMs are going to agentically do digitized paper shuffling and put digitized paper shufflers out of work. That is not the problem for my job. The issue is agentically from the ground up making the concept of digitized paper shuffling null and void. A relic of the past that can't compete in the economy.

Re: ChatGPT agent: bridging research and action

#334

Post by koolba »

dustingetz wrote: post #326
Aurornis wrote: post #127

Earlier quoted context omitted.

> how it normally takes him 4 to 8 hours to put together complicated, data-heavy reports. Now he fires off an agent request, goes to walk his dog, and comes back to a downloadable spreadsheet of dense data, which he pulls up and says "I think it got 98% of the information correct... This is where the AI hype bites people. A great use of AI in this situation would be to automate the collection and checking of data. Se…

98% sure each commit doesn’t corrupt the database, regress a customer feature, open a security vulnerability. 50 commits later … (which is like, one day for an agentic workflow)

It’s only a 64% chance of corruption after 50 such commits at a 98% success.

Re: ChatGPT agent: bridging research and action

#335

Post by wilg »

gitgud wrote: post #269
achrono wrote: post #134

Earlier quoted context omitted.

In typical SV style, this is just to throw it out there and let second order effects build up. At some point I expect OpenAI to simply form a partnership with LinkedIn and Amazon. In fact, I suspect LinkedIn might even create a new tier that you'd have to use if you want to use LinkedIn via OpenAI.

Why would platforms like LinkedIn want this? Bots have never been good for social media…

LinkedIn is probably the only social platform that would be improved by bots.

Re: ChatGPT agent: bridging research and action

#336

Post by nojs »

bredren wrote: post #236

This solves a big issue for existing CLI agents, which is session persistence for users working from their own machines. With claude code, you usually start it from your own local terminal. Then you have access to all the code bases and other context you need and can provide that to the AI. But when you shut your laptop, or have network availability changes the show stops. I've solved this somewhat on MacOS using the…

What tasks are you running that take more than a few minutes without intervention?

Re: ChatGPT agent: bridging research and action

#337

Post by com2kid »

bredren wrote: post #236

This solves a big issue for existing CLI agents, which is session persistence for users working from their own machines. With claude code, you usually start it from your own local terminal. Then you have access to all the code bases and other context you need and can provide that to the AI. But when you shut your laptop, or have network availability changes the show stops. I've solved this somewhat on MacOS using the…

Lightning.ai gives free CPU only dev boxes, I just run Claude code on one of those.

Re: ChatGPT agent: bridging research and action

#338

Post by candiddevmike »

NortySpock wrote: post #260
ivape wrote: post #196

Earlier quoted context omitted.

”The people who YOLO it with prompting cycles until the code passes tests and then submit a PR are causing problems almost as fast as they’re developing new features in non-trivial codebases.” This might as well be the new definition of “script kiddie”, and it’s the kids that are literally going to be the ones birthed into this lifestyle. The “craft” of programming may not be carried by these coming generations and p…

Oh come on, people have been writing code with bad, incomplete, flaky, or absent tests since automated testing was invented (possibly before). It's having a good, useful and reliable test suite that separates the sheep from the goats.* Would you rather play whack-a-mole with regressions and Heisenbugs, or ship features? * (Or you use some absurdly good programing language that is hard to get into knots with. I've bee…

Tests are just for the bugs you already know about

Re: ChatGPT agent: bridging research and action

#339

Post by xeonmc »

Aurornis wrote: post #127
twalkz wrote: post #30

The "spreadsheet" example video is kind of funny: guy talks about how it normally takes him 4 to 8 hours to put together complicated, data-heavy reports. Now he fires off an agent request, goes to walk his dog, and comes back to a downloadable spreadsheet of dense data, which he pulls up and says "I think it got 98% of the information correct... I just needed to copy / paste a few things. If it can do 90 - 95% of the…

> how it normally takes him 4 to 8 hours to put together complicated, data-heavy reports. Now he fires off an agent request, goes to walk his dog, and comes back to a downloadable spreadsheet of dense data, which he pulls up and says "I think it got 98% of the information correct... This is where the AI hype bites people. A great use of AI in this situation would be to automate the collection and checking of data. Se…

Or as I would like to put it, LLM outputs are essentially the Library of Babel. Yes, it contains all of the correct answers, but might as well be entirely useless.

Re: ChatGPT agent: bridging research and action

#340

Post by positron26 »

twalkz wrote: post #30

The "spreadsheet" example video is kind of funny: guy talks about how it normally takes him 4 to 8 hours to put together complicated, data-heavy reports. Now he fires off an agent request, goes to walk his dog, and comes back to a downloadable spreadsheet of dense data, which he pulls up and says "I think it got 98% of the information correct... I just needed to copy / paste a few things. If it can do 90 - 95% of the…

Distinguishing whether a problem is 0.02 ^ n for error or 0.98 ^ n for accuracy is emerging as an important skill.

Might explain why some people grind up a billion tokens trying to make code work only to have it get worse while others pick apart the bits of truth and quickly fill in their blind spots. The skillsets separating wheat from chaff are things like honest appreciation for corroboration, differentiating subjective from objective problems, and recognizing truth-preserving relationships. If you can find the 0.02 ^ n sub-problems, you can grind them down with AI and they will rapidly converge, leaving the 0.98 ^ n problems to focus human touch on.

Post reply on HN