Advancing AI Benchmarking with Game Arena
101 points
7 hours ago
| 14 comments
| blog.google
| HN
ofirpress
6 hours ago
[-]
This is a good way to benchmark models. We [the SWE-bench team] took the meta-version of this and implemented it as a new benchmark called CodeClash -

We have agents implement agents that play games against each other- so Claude isn't playing against GPT, but an agent written by Claude plays poker against an agent written by GPT, and this really tough task leads to very interesting findings on AI for coding.

https://codeclash.ai/

reply
63stack
5 hours ago
[-]
>this really tough task leads to very interesting findings on AI for coding

Are you going to share those with the class or?

reply
RobRivera
3 hours ago
[-]
reply
Instantnoodl
5 hours ago
[-]
Cool to see core war! I feel it's mostly forgotten by now. My dad is still playing it to this day though and even attends tournaments
reply
riku_iki
6 hours ago
[-]
Leaderboard looks very outdated..
reply
mohsen1
1 hour ago
[-]
Oh hey, I've been running Werewolves/Mafia games as benchmarks for a while now

https://mafia-arena.com

Gemini is consistently winning against top models

reply
ZeroCool2u
5 hours ago
[-]
I'd really like to see them add a complex open world fully physicalized game like Star Citizen (assuming the game itself is stable) with a single primary goal like accumulating currency as a measure of general autonomy and a proxy for how the model might behave in the real world given access to a bipedal robot.
reply
kenforthewin
2 hours ago
[-]
reply
iNic
3 hours ago
[-]
I feel uneasy about werewolf being included here. I don't want AI labs to actively try and make their LLMs deceptive!
reply
cv5005
6 hours ago
[-]
My personal threshold for AGI is when an AI can 'sit down' - it doesn't need to have robotic hands, but it needs to only use visual and audio inputs to make its moves - and complete a modern RPG or FPS single player game that it hasn't pre-trained on (it can train on older games).
reply
anematode
2 hours ago
[-]
Isn't this a bit too visual-centric? By this criterion Helen Keller, author of 14 books, would not be generally intelligent.

Ultimately I think it's impossible to define AGI. Maybe "I know it when I see it"—except everyone sees it at a different point (evidently).

reply
jamilton
1 hour ago
[-]
It could have hands that feel but no vision, I think they were getting at that they thought embodiment and playing games in the modality of humans, without thousands of hours of play to reach competency, would be an important milestone.
reply
bob1029
5 hours ago
[-]
reply
10xDev
6 hours ago
[-]
If AI can program, why does it matter if it can play Chess using CoT when it can program a Chess Engine instead? This applies to other domains as well.
reply
RivieraKid
4 hours ago
[-]
It can write a chess engine because it has read the code of a thousand of chess engines. This benchmark measures a different aspect of intelligence.

And as a poker player, I can say that this game is much more challenging for computers than chess, writing a program that can play poker really well and efficiently is an unsolved problem.

reply
10xDev
2 hours ago
[-]
The program doesn't need to be a solver. It can be anything that helps it.

It doesn't even need to be one tool but a series of tools.

reply
NitpickLawyer
4 hours ago
[-]
> If AI can program, why does it matter if it can play Chess using CoT when it can program a Chess Engine instead?

Heh, we really did come full circle on this! When chatgpt launched in dec22 one of the first things that people noticed is that it sucked at math. Like basic math 12 + 35 would trip it up. Then people "discovered" tool use, and added a calculator. And everyone was like "well, that's cheating, of course it can use a calculator, but look it can't do the simple addition logic"... And now here we are :)

reply
paxys
4 hours ago
[-]
IMO there's an expectation for baseline intelligence. I don't expect an "AGI" model to beat Magnus Carlsen out of the box but it should be able to do basic grade school level arithmetic and play chess at a complete beginner level without resorting to external tools.
reply
10xDev
3 hours ago
[-]
I'm not going to respond to everything but the key to my comment was "This applies to other domains as well." But people are limiting their imagination to the chess engine example given for chess. The tool or program (or even other neural networks that are available) can be literally anything for any task... Use your imagination.

Maybe we should just get rid of tedious benchmarks like chess altogether at this point that is leading people to think of how to limit AI as a way of keeping it a relevant benchmark rather than expanding on what is already there.

reply
Davidzheng
5 hours ago
[-]
They should be allowed to! In fact i think better benchmark would be to invent new games and test the models ability to allocate compute to minmax/alphazero new games in compute constraints
reply
simianwords
5 hours ago
[-]
Its the same reason we are asked to write exams without using calculators but the real world does have them.

How you work without calculators is a proxy for real world competency.

reply
10xDev
5 hours ago
[-]
Funny, you used probably the most useless form of benchmarking used on people as an example of measuring "competency" in the real world.
reply
doctorpangloss
5 hours ago
[-]
A lot of the insights of math come from knowing how to do things efficiently. That’s why the tests are timed. I don’t know, this is pretty basic pedagogy that you are choosing to grief.
reply
simianwords
5 hours ago
[-]
are you in favour of children using calculators in exams?
reply
10xDev
5 hours ago
[-]
It is a program. I need it to get task X done and I don't care how, whether it is strictly through CoT or with tools. There is no such thing as cheating in real work and no reason to handicap it. Just test the limits of what it can do with whatever means possible.

Trying to solve everything with CoT alone without utilising tools seems futile.

reply
simianwords
5 hours ago
[-]
you are not understanding. its a proxy for how well it does other things.
reply
10xDev
3 hours ago
[-]
A good proxy is knowing which tools to use to solve the problem. Not how to try and emulate how a human would play chess. That is pointless...
reply
CooCooCaCha
4 hours ago
[-]
CoT is upstream of building a chess engine.

Chess engines don’t grow on trees, they’re built by intelligent systems that can think, namely human brains.

Supposedly we want to build machines that can also think, not just regurgitate things created by human brains. That’s why testing CoT is important.

It’s not actually about chess, it’s about thinking and intelligence.

reply
mclau153
3 hours ago
[-]
Claude plays Pokemon Red
reply
tiahura
6 hours ago
[-]
How about nethack?
reply
tux3
3 hours ago
[-]
For reference for anyone who missed it, the 2021 NetHack challenge results: https://nethackchallenge.com/report.html

That was a whole half a decade ago, but back then deep learning AIs were defeated very badly by handcrafted scripts. Even the best bot in the neural net category was actual a symbolic script/neural net hybrid.

reply
eamag
6 hours ago
[-]
Curious why they decided to curate poker hands instead of a normal poker
reply
qsort
6 hours ago
[-]
Poker has very high variance, you'd need several hundred thousand hands to confidently say who's better. Also, you probably want to precompute the GTO-optimal play for benchmarking purposes.
reply
johndhi
6 hours ago
[-]
But can't computers play several hundred thousand poker hands easily in a couple of hours ?
reply
eamag
6 hours ago
[-]
But now because the hands are so strong we don't see any folds
reply
simianwords
5 hours ago
[-]
Gemini tops all benchmarks but when it comes to real world usage it is genuinely unusable
reply
CuriouslyC
4 hours ago
[-]
It's legit good at visual stuff. It's not just a great agent and does some weird stuff sometimes.
reply
goniszewski
5 hours ago
[-]
It’s not that bad. I’ve been using 3 Pro for some time now and I’m quite happy with how it works. Best paired with Opus and Codex, like most models, but it’s solid as a full-stack buddy.
reply
bennyfreshness
5 hours ago
[-]
Wow. I'm generally in the AI maximalist camp. But adding Werewolf feels dangerous to me. Anyone who's played knows lying, deceipt, and manipulation is often key to winning. We really want models climbing this benchmark?
reply
rustyhancock
4 hours ago
[-]
Oddly in the highlighted game I watched the werewolf simply gives up in the last round and says I'm the werewolf well-done... Vote me.

Bizarre.

reply
minihat
1 hour ago
[-]
This is a legitimate strategy for the werewolf, no?
reply
bilekas
5 hours ago
[-]
Good question, but who's going to stop them?

AI already has a very creative imagination for role play so this just adds extra to their arsenal.

reply
PunchyHamster
4 hours ago
[-]
confidently and charismatically lying to clueless users has been one of fundaments of AI adoption
reply
chaostheory
6 hours ago
[-]
Anecdotal data point, but recently I’ve found Gemini to perform better than ChatGPT when it came to intent analysis.
reply
PunchyHamster
4 hours ago
[-]
making models target benchmark about being good at lying and getting away with it (werewolf) is certainly an interesting choice
reply