The general conceit of this article, which is something that many frontier labs seem to be beginning to realize, is that the average human is no longer smart enough to provide sufficient signal to improve AI models.
No, it's that the average unpaid human doesn't care to read closely enough to provide signal to improve AI models. Not that they couldn't if they put in even the slightest amount of effort.
Firstly, paying is not at all the correct incentive for the desired outcome. When the incentive is payment, people will optimize for maximum payout not for the quality goals of the system.
Secondly, it doesn't fix stupidity. A participant who earnestly takes the quality goals of the system to heart instead of focusing on maximizing their take (thus, obviously stupid) will still make bad classifications due to that reason.
> Firstly, paying is not at all the correct incentive for the desired outcome. When the incentive is payment, people will optimize for maximum payout not for the quality goals of the system.
1. I would expect any paid arrangement to include a quality-control mechanism. With the possible exception of if it was designed from scratch by complete ignoramuses.
I'm being (mostly) serious, suppose you're a stuffed ahort trying to boost your valuation, how can you work out who's smart enough to train your LLM? (Never mind how to get them to work for you!)
I do a lot of human evaluations. Lots of Bayesian / statistical models that can infer rater quality without ground truth labels. The other thing about preference data you have to worry about (which this article gets at) is: preferences of _who_? Human raters are a significantly biased population of people, different ages, genders, religions, cultures, etc all inform preferences. Lots of work being done to leverage and model this.
Then for LMArena there is the host of other biases / construct validity: people are easily fooled, even PhD experts; in many cases it’s easier for a model to learn how to persuade than actually learn the right answers.
But a lot of dismissive comments as if frontier labs don’t know this, they have some of the best talent in the world. They aren’t perfect but they in a large sene know what they’re doing and what the tradeoffs of various approaches are.
Human annotations are an absolute nightmare for quality which is why coding agents are so nice: they’re verifiable and so you can train them in a way closer to e.g. alphago without the ceiling of human performance
Sure, on the surface judging the judge is just as hard as being the judge
But at least the two examples of judging AI provided in the article can be solved by any moron by expending enough effort. Any moron can tell you what Dorothy says to Toto when entering Oz by just watching the first thirty minutes of the movie. And while validating answer B in the pan question takes some ninth-grade math (or a short trip to wikipedia), figuring out that a nine inch diameter circle is in fact not the same area as a 9x13 inch square is not rocket science. And with a bit of craft paper you could evaluate both answers even without math knowledge
So the short answer is: with effort. You spend lots of effort on finding a good evaluator, so the evaluator can judge the LLM for you. Or take "average humans" and force them to spend more effort on evaluating each answer
Popularity has never been a meaningful signal of quality, no matter how many tech companies try to make it so, with their star ratings, up/down voting, and crowdsourcing schemes.
Different strokes for different folks: I mean who is to say if Bleach or Backstabbed in a Backwater Dungeon: My Trusted Companions Tried to Kill Me, but Thanks to the Gift of an Unlimited Gacha I Got LVL 9999 Friends and Am Out for Revenge on My Former Party Members and the World is better?
Yep, it's like getting a commoner from the street evaluate a literature PhD in their native language. Sure, both know the language, but the depth difference of a specialist vs a generalist is too large. And neither we can't use AI to automatically evaluate this literature genius because real AI doesn't exist (yet), hence the programs can't understand the contents of text they output or input. Whoops. :)
The average human is a moron you wouldn't trust to watch your hamster. If you watched them outside of the narrow range of tasks they have been trained to perform by rote you would probably conclude they should qualify for benefits by virtue of mental disability.
We give them WAY too much credit by watching mostly the things they have been trained specifically to do and pretending this indicates a general mental competence that just doesn't exist.
I kinda assumed they wouldn't need any money because AI companies give them free credits to evaluate the models, and users ask questions and rate for free because they get to use decent AI models at no cost...
Beyond that there is coding up a web page, which as we all know can be vibe coded in a few hours...
Oh my goodness yes, I almost missed it that the text is (mostly?) AI written. That said I agree that LMArena elo scores are pushing models in the wrong direction. They move more towards McDonald's than quality food.
How can you tell? (honest question, I really can't)
The article makes strong points, includes real data and quotes, shows proof of work (sampling 100 Q&A), so does that even matter at this point? This doesn't feel like "slop" to me at all.
Yea I also didn't think this was written by ai, it sounded human enough to me. It's kind of a bummer that there's all these patterns that LLM's follow in their output that cause people to have a knee jerk reaction and instantly call it ai slop. I know there is a ton of ai garbage out there these days, but I really couldn't tell with this article.
The text definitely the "jump from dramatic crescendo to dramatic crescendo" quality of certain LLM texts. If you read closely, it also has adjective choice that's more for dramatic than appropriate to the circumstances involves (a quality of LLM texts it also helpfully explains).
I don't know if this proves it's an LLM text or whether that style is simply spilling out everywhere.
> They're not reading carefully. They're not fact-checking, or even trying.
It’s not how I do, and I suppose how many people do. I specifically ask questions related to niche subjects that I know perfectly well and that is very easy for me to spot mistakes.
The first time I used it, that’s what came naturally to my mind. I believe it’s the same for others.
Yeah, that quote just reads like the typical “everyone is an idiot except me” attitude that pervades the tech world.
Of course people visiting a website specifically designed for evaluating LLMs do try all kinds of specific things to specifically test for weaknesses. There may be users who just click on the response with more emojis, but I strongly doubt they are the majority on that particular site.
When they released GPT-4.5, it was miles ahead of others when it comes to its linguistic skills and insight. Yet, it was never at top of the arena - it felt that not everone was able to appreciate the edge.
>Would you trust a medical system measured by: which doctor would the average Internet user vote for?
Yes, the system desperately needs this. Many doctors malpractice for DECADES.
I would absolutely seek to, damn, even pay good money to, be able to talk with a doctor's previous patients, particularly if they're going to perform a life-changing procedure on me.
Raw score is often quite frankly crap. It's often still easy to surface the negative reviews and since people don't at least at present fake those you can find out what they didn't like about a product. If a given products critics are only those whining about something irrelevant, not meaningful to your use case, or acceptable to you and it overall appears to meet spec you are often golden.
this argument is also broadly true about the quality and correctness of posts on any vote-based discussion board
> Why is LMArena so easy to game? The answer is structural.
> The system is fully open to the Internet. LMArena is built on unpaid labor from uncontrolled volunteers.
also all user's votes count equally, bu not all users have equal knowledge.
As long as users are better than 50% accurate, it shouldn't matter if they're experts or not. That being said, it's difficult to measure user accuracy in this case without running into circular reasoning.
True and what you can realize/read between the lines is something deeper.
LLMs are fallible.
Humans are fallible.
LLMs improve (and improve fast).
Humans do not (overall, ie. "group of N experts in X", "N random internet people").
All those "turing tests" will start flipping.
Today it's "N random internet humans" score too low on those benchmarks, tomorrow it'll be "group of N expert humans in X" score too low.
> Being verbose. Longer responses look more authoritative!
I know we can solve this in ordinary tasks just using prompt but that's really annoying. Sometimes I just want a yes or no answer and then I get a phd thesis in the matter.
Aside from Meta is there any reason to think the big AI labs are still using LMArena data for training? The weaknesses are well understood and with the shift to RL there are so many better ways to design a reward function.
Is there a reason wrong data isn't considered more broadly in its context as still valuable?
Shouldn't the model effectively 1. learn to complete the incorrect thing and 2. learn the context that it's correct and incorrect? In this case the context being lazy LMArena users. And presumably, in the future, poorly filtered training data.
We seem to be able to read incorrect things and not be corrupted (well, theoretically). It's not ideal, but it seems an important component to intellectual resilience.
It seems like the model knowing the data is LMArena, or some type of un-trusted, would be sufficient to shift the prior to a reasonable place.
maybe it would work if they could encourage end users to be rigorous? (ie, detect if they have the capability to rate well and then reward them when they do by comparing them against other highly rated raters of the same phenotype)
> Voilà: bold text, emojis, and plenty of sycophancy – every trick in the LMArena playbook! – to avoid answering the question it was asked.
This is hard to swallow.
I don't believe a single word this article says. Apparently the "real author" (the human being who wrote the original prompt to generate this article) only intend to use this article to generate clicks and engagement but don't care at all about what's in there.
The general conceit of this article, which is something that many frontier labs seem to be beginning to realize, is that the average human is no longer smart enough to provide sufficient signal to improve AI models.
No, it's that the average unpaid human doesn't care to read closely enough to provide signal to improve AI models. Not that they couldn't if they put in even the slightest amount of effort.
Firstly, paying is not at all the correct incentive for the desired outcome. When the incentive is payment, people will optimize for maximum payout not for the quality goals of the system.
Secondly, it doesn't fix stupidity. A participant who earnestly takes the quality goals of the system to heart instead of focusing on maximizing their take (thus, obviously stupid) will still make bad classifications due to that reason.
> Firstly, paying is not at all the correct incentive for the desired outcome. When the incentive is payment, people will optimize for maximum payout not for the quality goals of the system.
1. I would expect any paid arrangement to include a quality-control mechanism. With the possible exception of if it was designed from scratch by complete ignoramuses.
2. Do you have a proposal for a better incentive?
Why would an unpaid human want to do that?
Therein lies the problem.
Exactly — they wouldn't.
But when you're a moron how can you distinguish?
I'm being (mostly) serious, suppose you're a stuffed ahort trying to boost your valuation, how can you work out who's smart enough to train your LLM? (Never mind how to get them to work for you!)
I do a lot of human evaluations. Lots of Bayesian / statistical models that can infer rater quality without ground truth labels. The other thing about preference data you have to worry about (which this article gets at) is: preferences of _who_? Human raters are a significantly biased population of people, different ages, genders, religions, cultures, etc all inform preferences. Lots of work being done to leverage and model this.
Then for LMArena there is the host of other biases / construct validity: people are easily fooled, even PhD experts; in many cases it’s easier for a model to learn how to persuade than actually learn the right answers.
But a lot of dismissive comments as if frontier labs don’t know this, they have some of the best talent in the world. They aren’t perfect but they in a large sene know what they’re doing and what the tradeoffs of various approaches are.
Human annotations are an absolute nightmare for quality which is why coding agents are so nice: they’re verifiable and so you can train them in a way closer to e.g. alphago without the ceiling of human performance
> in many cases it’s easier for a model to learn how to persuade than actually learn the right answers
So we should expect the models to eventually tend toward the same behaviors that politicians exhibit?
Maybe a happy to deceive marketing/sales role would be more accurate.
100% (am a Bayesian statistician).
Isn’t it fascinating how it comes down to quality of judgement (and the descriptions thereof)?
We need an LMArena rated by experts.
As a statistician, do you you think you could, given access to the data, identify the subset of LMArena users that are experts?
they always know, they just have non-AGI incentive and asymetric upside to play along...
Sure, on the surface judging the judge is just as hard as being the judge
But at least the two examples of judging AI provided in the article can be solved by any moron by expending enough effort. Any moron can tell you what Dorothy says to Toto when entering Oz by just watching the first thirty minutes of the movie. And while validating answer B in the pan question takes some ninth-grade math (or a short trip to wikipedia), figuring out that a nine inch diameter circle is in fact not the same area as a 9x13 inch square is not rocket science. And with a bit of craft paper you could evaluate both answers even without math knowledge
So the short answer is: with effort. You spend lots of effort on finding a good evaluator, so the evaluator can judge the LLM for you. Or take "average humans" and force them to spend more effort on evaluating each answer
Maybe you need to have people rate others ratings to remove at least the worst idiots.
that’s why Mercor is worth 2billion
Popularity has never been a meaningful signal of quality, no matter how many tech companies try to make it so, with their star ratings, up/down voting, and crowdsourcing schemes.
Different strokes for different folks: I mean who is to say if Bleach or Backstabbed in a Backwater Dungeon: My Trusted Companions Tried to Kill Me, but Thanks to the Gift of an Unlimited Gacha I Got LVL 9999 Friends and Am Out for Revenge on My Former Party Members and the World is better?
It is glaringly obvious that the average human is not smart enough to the level hat their decision making should be replicated and adopted at scale.
People hold falsehoods to be true, and cannot calculate a 10% tip.
Yep, it's like getting a commoner from the street evaluate a literature PhD in their native language. Sure, both know the language, but the depth difference of a specialist vs a generalist is too large. And neither we can't use AI to automatically evaluate this literature genius because real AI doesn't exist (yet), hence the programs can't understand the contents of text they output or input. Whoops. :)
The average human is a moron you wouldn't trust to watch your hamster. If you watched them outside of the narrow range of tasks they have been trained to perform by rote you would probably conclude they should qualify for benefits by virtue of mental disability.
We give them WAY too much credit by watching mostly the things they have been trained specifically to do and pretending this indicates a general mental competence that just doesn't exist.
If these frontier models were open source, the market of downstream consumers would figure out how to optimize them.
By being closed, they'll never be optimal.
They need to spend money on actual experts to curate their data to improve.
Instead, finance bros are convinced by the argument that number goes up.
Is that not exactly what https://www.mercor.com/ does?
Wait you know that frontier labs do actually do this right?
Sometimes it feels like:
AI will make it cheaper, faster, better, no problem. You can eat the cake now and save it for later.> It's past time for LMArena people to sit down and have some thorough reflection on whether it is still worth running at all
They've raised about $250 million, so I don't see that happening anytime soon.
I kinda assumed they wouldn't need any money because AI companies give them free credits to evaluate the models, and users ask questions and rate for free because they get to use decent AI models at no cost...
Beyond that there is coding up a web page, which as we all know can be vibe coded in a few hours...
What else is there to spend money on?
i asked them this in my interview. tldr they subsidize all inference on their platform https://www.youtube.com/watch?v=NBnOk0Uy9ig&t=70s
They don't need to spend extensively for tokens, but they gain extensively from charging for access once they've become an established player.
But the question was: what do they need $250m for?
"so that we can move even faster to build new features and improve our product experience for all our users" https://news.lmarena.ai/series-a/
everyone needs $250mil :)
There's something deeply ironic about this being written by AI. Baitception, even.
"The Brutal Choice"
Is there an established name for this LLMism?
I don't need a "Reality Check" or a "Hard Truth". The thought can be concluded without this performative honesty nonsense or the emotive hyperbole.
This probably grates me more than any other.
Oh my goodness yes, I almost missed it that the text is (mostly?) AI written. That said I agree that LMArena elo scores are pushing models in the wrong direction. They move more towards McDonald's than quality food.
This was my first thought as well
How can you tell? (honest question, I really can't)
The article makes strong points, includes real data and quotes, shows proof of work (sampling 100 Q&A), so does that even matter at this point? This doesn't feel like "slop" to me at all.
Yea I also didn't think this was written by ai, it sounded human enough to me. It's kind of a bummer that there's all these patterns that LLM's follow in their output that cause people to have a knee jerk reaction and instantly call it ai slop. I know there is a ton of ai garbage out there these days, but I really couldn't tell with this article.
The text definitely the "jump from dramatic crescendo to dramatic crescendo" quality of certain LLM texts. If you read closely, it also has adjective choice that's more for dramatic than appropriate to the circumstances involves (a quality of LLM texts it also helpfully explains).
I don't know if this proves it's an LLM text or whether that style is simply spilling out everywhere.
> They're not reading carefully. They're not fact-checking, or even trying.
It’s not how I do, and I suppose how many people do. I specifically ask questions related to niche subjects that I know perfectly well and that is very easy for me to spot mistakes.
The first time I used it, that’s what came naturally to my mind. I believe it’s the same for others.
Unfortunately I don't think there's any reason to assume that you're a representative sample of LMArena users.
Yeah, that quote just reads like the typical “everyone is an idiot except me” attitude that pervades the tech world.
Of course people visiting a website specifically designed for evaluating LLMs do try all kinds of specific things to specifically test for weaknesses. There may be users who just click on the response with more emojis, but I strongly doubt they are the majority on that particular site.
When they released GPT-4.5, it was miles ahead of others when it comes to its linguistic skills and insight. Yet, it was never at top of the arena - it felt that not everone was able to appreciate the edge.
4.5 was easily the best conversationalist I've seen. Not as powerful as modern ones but something about HOW it talked felt inherently smart.
I miss that one, is 5 any better? I switched to claude before it launched.
> something about HOW it talked felt inherently smart
The thing was huge. They were training the thing to be GPT5, before they figured out their userbase to too large to be served something that big.
No replacement for displacement, except applied to LLMs and raw parameter count.
>Would you trust a medical system measured by: which doctor would the average Internet user vote for?
Yes, the system desperately needs this. Many doctors malpractice for DECADES.
I would absolutely seek to, damn, even pay good money to, be able to talk with a doctor's previous patients, particularly if they're going to perform a life-changing procedure on me.
Doctors would also pay good money for votes, so I'm not sure that would fix anything.
Raw score is often quite frankly crap. It's often still easy to surface the negative reviews and since people don't at least at present fake those you can find out what they didn't like about a product. If a given products critics are only those whining about something irrelevant, not meaningful to your use case, or acceptable to you and it overall appears to meet spec you are often golden.
Seems like they just raised 150m at 1.7B valuation. Crazy.
Source: https://techcrunch.com/2026/01/06/lmarena-lands-1-7b-valuati...
Who? LMArena? That's actually crazy.
Are they selling:
A. model improvement tests, suites, and benchmarks
B. data on competitors' evals
C. test answer keys
D. alpha to VC firms
E. all of the above
???
Apparently they are selling model evaluations, powered by their volunteer users.
They're selling "I'm an AI investor" stickers to show off at the next family reunion
I'm taking the Red Cross public next. With the price of healthcare these days my earnings projections are uber-extreme.
From https://lmarena.ai/how-it-works:
> In battle mode, you'll be served 2 anonymous models. Dig into the responses and decide which answer best fits your needs.
It's not a given that someone's needs are "factual accuracy". Maybe they're after entertainment, or winning an argument.
Any metric that can be targeted can be gamed
Then target it with metrics worth solving[1].
1. Ex https://mppbench.com/
But that seems to be measuring "superintelligence" rather than just AI, no?
If the metric is a latent variable summarizing subjective judgements, yes.
this argument is also broadly true about the quality and correctness of posts on any vote-based discussion board
> Why is LMArena so easy to game? The answer is structural. > The system is fully open to the Internet. LMArena is built on unpaid labor from uncontrolled volunteers.
also all user's votes count equally, bu not all users have equal knowledge.
As long as users are better than 50% accurate, it shouldn't matter if they're experts or not. That being said, it's difficult to measure user accuracy in this case without running into circular reasoning.
True and what you can realize/read between the lines is something deeper.
LLMs are fallible. Humans are fallible. LLMs improve (and improve fast). Humans do not (overall, ie. "group of N experts in X", "N random internet people").
All those "turing tests" will start flipping.
Today it's "N random internet humans" score too low on those benchmarks, tomorrow it'll be "group of N expert humans in X" score too low.
The average person is dumber than an LLM in terms of having a grasp on the facts, and basic arithmetic.
A voting system open to the public is completely screwed even if somehow its incentives are optimized toward strongly encouraging ideal behavior.
> Being verbose. Longer responses look more authoritative!
I know we can solve this in ordinary tasks just using prompt but that's really annoying. Sometimes I just want a yes or no answer and then I get a phd thesis in the matter.
Couldn't "The Wisdom of Crowds" help with this?
Maybe if they started ranking the answers on a 1-10 range, allowing people to specify graduations of correctness/wrongness, then the crowd would work?
https://en.wikipedia.org/wiki/The_Wisdom_of_Crowds
AI is a cancer on humanity
When the Meta cheating scandal happened I was surprised how little of the attention was on this.
Meta "cheated" on lmarena not by using a smarter model but by using one that was more verbose and friendly with excessive emojis.
Since AI is itself a cancer, maybe this is good? The cancer of my cancer is my chemo.
Is there any reason to believe LMArena isn't botted by the people releasing these models?
> It's like going to the grocery store and buying tabloids, pretending they're scientific journals.
This is pure gold. I've always found this approach of evals on a moving-target via consensus broken.
Aside from Meta is there any reason to think the big AI labs are still using LMArena data for training? The weaknesses are well understood and with the shift to RL there are so many better ways to design a reward function.
I don't think anyone has ever used it as training. But yes labs still do seem to target it as goal (which is a different thing).
Such as?
Is there a reason wrong data isn't considered more broadly in its context as still valuable?
Shouldn't the model effectively 1. learn to complete the incorrect thing and 2. learn the context that it's correct and incorrect? In this case the context being lazy LMArena users. And presumably, in the future, poorly filtered training data.
We seem to be able to read incorrect things and not be corrupted (well, theoretically). It's not ideal, but it seems an important component to intellectual resilience.
It seems like the model knowing the data is LMArena, or some type of un-trusted, would be sufficient to shift the prior to a reasonable place.
We need a service that ranks AI model ranking services. Maybe powered by AI instead of humans?
Just look at Open(ugh)Router. That's a good, though not fully accurate, view of where dollars are going.
It'd be nice if it were actually open and we could inspect all the statistics.
maybe it would work if they could encourage end users to be rigorous? (ie, detect if they have the capability to rate well and then reward them when they do by comparing them against other highly rated raters of the same phenotype)
> What actually happens: random Internet users spend two seconds skimming, then click their favorite.
> They're not reading carefully. They're not fact-checking, or even trying.
Uhhh, how was that established?
> Voilà: bold text, emojis, and plenty of sycophancy – every trick in the LMArena playbook! – to avoid answering the question it was asked.
This is hard to swallow.
I don't believe a single word this article says. Apparently the "real author" (the human being who wrote the original prompt to generate this article) only intend to use this article to generate clicks and engagement but don't care at all about what's in there.
and AI is a cancer on humanity... this article is clearly LLM written too.
Poison Fountain: https://rnsaffn.com/poison3/