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

Study identifies weaknesses in how AI systems are evaluated

oii.ox.ac.uk

Post reply on HN
Print view View original

91–100 of 204 posts

Re: Study identifies weaknesses in how AI systems are evaluated

#91

Post by simonw »

riskable wrote: post #85

We should make a collective git repo full of every kind of annoying bug we (expert developers) can think of. Then use that to benchmark LLMs. Someone want to start? I've got a Yjs/CRDT collaborative editing bug that took like a week and a half of attempts with Claude Code (Sonnet 4.5), GPT5-codex (medium), and GLM-4.6 many, many attempts to figure out. Even then they didn't really get it... Just came up with a succes…

Have you tried the Playwright libraries? Not the MCP, instead telling Claude Code to use the Node.js or Python Playwright libraries directly. I have had some really good results for this for gnarly frontend challenges.

Re: Study identifies weaknesses in how AI systems are evaluated

#92

Post by cortesoft »

riskable wrote: post #85

We should make a collective git repo full of every kind of annoying bug we (expert developers) can think of. Then use that to benchmark LLMs. Someone want to start? I've got a Yjs/CRDT collaborative editing bug that took like a week and a half of attempts with Claude Code (Sonnet 4.5), GPT5-codex (medium), and GLM-4.6 many, many attempts to figure out. Even then they didn't really get it... Just came up with a succes…

It would be pretty easy to over fit the results with a static set of tests

Re: Study identifies weaknesses in how AI systems are evaluated

#93

Post by bogwog »

riskable wrote: post #85

We should make a collective git repo full of every kind of annoying bug we (expert developers) can think of. Then use that to benchmark LLMs. Someone want to start? I've got a Yjs/CRDT collaborative editing bug that took like a week and a half of attempts with Claude Code (Sonnet 4.5), GPT5-codex (medium), and GLM-4.6 many, many attempts to figure out. Even then they didn't really get it... Just came up with a succes…

I can't tell how much of this is sarcasm

> we (expert developers) ...

> took like a week and a half of attempts with Claude Code ...

What kind of expert developer wastes that much time prompting a bunch of different LLMs to end up with a workaround, instead of actually debugging and fixing the bug themselves?

Re: Study identifies weaknesses in how AI systems are evaluated

#94

Post by jstummbillig »

Benchmarks are like SAT scores. Can they guarantee you'll be great at your future job? No, but we are still roughly okay with what they signify. Clearly LLMs are getting better in meaningful ways, and benchmarks correlate with that to some extend.

Re: Study identifies weaknesses in how AI systems are evaluated

#95

Post by kvirani »

bogwog wrote: post #93
riskable wrote: post #85

We should make a collective git repo full of every kind of annoying bug we (expert developers) can think of. Then use that to benchmark LLMs. Someone want to start? I've got a Yjs/CRDT collaborative editing bug that took like a week and a half of attempts with Claude Code (Sonnet 4.5), GPT5-codex (medium), and GLM-4.6 many, many attempts to figure out. Even then they didn't really get it... Just came up with a succes…

I can't tell how much of this is sarcasm > we (expert developers) ... > took like a week and a half of attempts with Claude Code ... What kind of expert developer wastes that much time prompting a bunch of different LLMs to end up with a workaround, instead of actually debugging and fixing the bug themselves?

Fair question but I think the tone of this is a bit abrasive towards the poster, and unnecessarily so.

Re: Study identifies weaknesses in how AI systems are evaluated

#96

Post by kvirani »

simonw wrote: post #91
riskable wrote: post #85

We should make a collective git repo full of every kind of annoying bug we (expert developers) can think of. Then use that to benchmark LLMs. Someone want to start? I've got a Yjs/CRDT collaborative editing bug that took like a week and a half of attempts with Claude Code (Sonnet 4.5), GPT5-codex (medium), and GLM-4.6 many, many attempts to figure out. Even then they didn't really get it... Just came up with a succes…

Have you tried the Playwright libraries? Not the MCP, instead telling Claude Code to use the Node.js or Python Playwright libraries directly. I have had some really good results for this for gnarly frontend challenges.

Curious why not the MCP? I use that

Re: Study identifies weaknesses in how AI systems are evaluated

#97

Post by DrewADesign »

benjiro wrote: post #70
luke0016 wrote: post #39

Earlier quoted context omitted.

> reasoning ability should enable them to handle numbers of arbitrary size, just as it enables humans to do so, given some pencil and paper. Or given a calculator. Which it's running on. Which it in some sense is . There's something deeply ironic about the fact that we have an "AI" running on the most technologically advanced calculator in the history of mankind and...it can't do basic math.

Thing is, a LLM is nothing but a prediction algorithm based upon what it trained. So it missing basic calculator functionality is a given. This is why tool usage is more and more a thing for LLMs. So that the LLM can from itself use a calculator for the actual math parts it needs. Thus increasing accuracy ...

If they were selling LLMs as “LLMs” instead of magic code-writing, answer-giving PhD replacements, the lack of basic arithmetic capability would be a given… but they aren’t. Judging a paid service using their own implied claims is perfectly reasonable.

Re: Study identifies weaknesses in how AI systems are evaluated

#98

Post by botro »

calpaterson wrote: post #3

Definitely one of the weaker areas in the current LLM boom. Comparing models, or even different versions of the same model, is a pseudo-scientific mess. I'm still using https://lmarena.ai/leaderboard . Perhaps there is something better and someone will pipe up to tell me about it. But we use LLMs at work and have unexplainable variations between them. And when we get a prompt working reliably on one model, we often h…

This is something I've stuggled with for my site, I made https://aimodelreview.com/ to compare the outputs of LLMs over a variety of prompts and categories, allowing a side by side comparison between them. I ran each prompt 4 times for each model with different temperature values available as a toggles.

My thinking was to just make the responses available to users and let them see how models perform. But from some feedback, turns out users don't want to have to evaluate the answers and would rather see a leaderboard and rankings.

The scalable solution to that would be LLM as judge that some benchmarks already use, but that just feels wrong to me.

LM Arena tries to solve this with the crowd sourced solution, but I think the right method would have to be domain expert human reviewers, so like Wirecutter VS IMDb, but that is expensive to pull off.

Re: Study identifies weaknesses in how AI systems are evaluated

#99

Post by s900mhz »

kvirani wrote: post #96
simonw wrote: post #91

Earlier quoted context omitted.

Have you tried the Playwright libraries? Not the MCP, instead telling Claude Code to use the Node.js or Python Playwright libraries directly. I have had some really good results for this for gnarly frontend challenges.

Curious why not the MCP? I use that

When I have a bug I’m iterating on it’s much easier and faster to have it write out the playwright script. That way it does not have to waste time or tokens performing the same actions over and over again.

Think of it as TDD.

Re: Study identifies weaknesses in how AI systems are evaluated

#100

Post by DrewADesign »

yawnxyz wrote: post #49
SurceBeats wrote: post #30

Benchmarks optimize for fundraising, not users. The gap between "state of the art" and "previous gen" keeps shrinking in real-world use, but investors still write checks based on decimal points in test scores.

we try to make benchmarks for users, but it's like that 20% article - different people want different 20% and you just end up adding "features" and whackamoling the different kinds of 20% if a single benchmark could be a universal truth, and it was easy to figure out how to do it, everyone would love that.. but that's why we're in the state we're in right now

The problem isn’t with the benchmarks (or the models, for that matter) it’s their being used to prop up the indefensible product marketing claims made by people frantically justifying asking for more dump trucks of thousand-dollar bills to replace the ones they just burned through in a few months.
Post reply on HN