Rendered at 06:44:27 GMT+0000 (Coordinated Universal Time) with Cloudflare Workers.
mey 11 hours ago [-]
A shared blind spot as noted at the bottom needs to be considered more often. In my day job, most of my coordination with others and now LLMs, is clarifying context and requirements. Claude is very happy to make assertions without the full picture in my experience, even when I give it as much context as I can.
lhk931122 3 hours ago [-]
Since the baseline LLM architecture is similar with each other, maybe already there are relations between their each opinion. Of course, this is an assumption and the explicit training seems effective in this case. But, I'm curious whether the approach would be effective in other cases (in terms of generalization?)
qarl 12 hours ago [-]
While this is absolutely true - I'd hesitate to discount using similar agents for checking each other. Two agents will almost never hallucinate in the same way, regardless of their weights - and by having a second one (with a different context) check almost entirely eliminates the problem.
Frieren 11 hours ago [-]
> and by having a second one (with a different context) check almost entirely eliminates the problem.
You solved one of the largest problems with current LLMs. How is it possible that nobody tried that before?
Because they do. There are already LLMs checking outputs of other LLMs, the bullshit answers that you see are the results of failures on that checks. If you remove all checks LLMs will create hallucinations even more often.
qarl 9 hours ago [-]
I'm sorry you're so upset, but it's true. Try it for yourself - get your LLM to hallucinate something, and then paste that text into another chat window and ask it to verify the result for you.
kmoser 3 hours ago [-]
By design, even the same LLM, when asked the same question multiple times, will almost never hallucinate in the same way. By that (flawed) logic, you could have the same LLM check itself.
emodendroket 12 hours ago [-]
It depends what we're judging, doesn't it? If it's "is the formatting in this document compliant with our standards?" I think it's reasonable. If it's like, life-altering if it's wrong I'm less sanguine.
Joel_Mckay 11 hours ago [-]
They have already shown algorithmic discrimination in predicting recidivism for brown people, as they are nonsensically overrepresented in the statistical data of US prison populations.
Folks should sue in a class-action lawsuit, any legal firm worth their beautiful walnut desks would seriously be happy take on that constitutionally backed mission. =3
suburban_strike 9 hours ago [-]
> brown people, as they are nonsensically overrepresented in the statistical data of US prison populations
"Nonsensical?" They commit violent crimes, they get prosecuted for said violent crimes, and are serving prison sentences for those crimes. The algorithm picks up on this trend using the same logic that insurance actuaries use, which has also been largely neutered by critical theory.
What even is the argument here-- they're all innocent? Cops are ignoring piles of dead white people and their white murderers to only go patrol brown neighborhoods? We both know neither claim is true. The usual complaint is that cops avoid their neighborhoods and/or are lazy in investigating the crimes they report. The idea of overpolicing has always been a Marxist double-bind...nonsensical, I daresay.
qarl 8 hours ago [-]
Oh. Yes. Sure. If there is an error in the training data then they will obviously cite that error.
I'm mostly talking about random coding errors.
piyh 9 hours ago [-]
No modern LLM can tell how many eyes the magic card Pit Imp has. They all say 2. This is across all reasoning levels and paid Gemini, Claude, and GPT (Sol)
Drawing a line red to split up the image then has them answer correctly.
Their failure modes are highly correlated.
qarl 8 hours ago [-]
Yes - those failures (like strawberry) are. And those failures are very rare, which is why you had to reach for the Pit Imp MTG card, which I had to Google to understand your point.
knollimar 6 hours ago [-]
[dead]
someguynamedq 4 hours ago [-]
> Two agents will almost never hallucinate in the same way, regardless of their weights
Citation needed
qarl 2 hours ago [-]
Personal experience using agents and seeing this happen frequently.
Try it yourself. Get one to hallucinate, then paste that text into a new window and ask it to verify the facts.
EDIT:
Also - Cohen, Hamri, Geva & Globerson, "LM vs LM: Detecting Factual Errors via Cross Examination": Cross-examination "detects over 70% of the incorrect claims while maintaining a high precision of >80%".
So 70% for ANY error, not just hallucinations.
bryzaguy 13 hours ago [-]
They would all agree raspberry has two Rs
Joel_Mckay 12 hours ago [-]
But still refuse to answer "How many strings does a bass play with in water?" correctly, perhaps the chat monitors in the third world data entry centers will manually patch the nonsense for a more rational answer someday. lol =3
wbobeirne 11 hours ago [-]
This was the first time I'd heard that gotcha question. I just threw it at Opus 5:
None — a bass in water is a fish, and fish are notoriously bad at music.
The instrument version plays four strings as standard (five and six-string basses exist for players who want to go lower or higher), and it prefers to stay dry.
Seems like a pretty good answer to me!
Joel_Mckay 11 hours ago [-]
Indeed, giving a definite answer to an ambiguous nonsense question is still incorrect.
A fish can play with as many strings as it finds, but only one when on a hook. Yet this too is an incorrect answer, as it again ignores the ambiguity in the phrasing. =3
aesthesia 10 hours ago [-]
This feels like an xkcd 169 situation, to be honest.
Joel_Mckay 10 hours ago [-]
It was actually a trivial allusion to a rather old poetic parable, and highlights a foundational flaw in LLM inference model statistical salience.
If a LLM based chat bot does ever answer it correctly, than you know with a fair degree of certainty it was content moderators stepping into the chat. Have a wonderful day. =3
OK, I still have no idea what you actually think the "correct" answer is.
quietbritishjim 9 hours ago [-]
I think they want the correct answer to be "your question doesn't really make sense, so I'm not going to answer it". (But I also think Opus's answer is better.)
Joel_Mckay 8 hours ago [-]
All the answers I've seen so far are assuredly not inaccurate (the elephants trunk is like a snake), but never fully correct (an elephant is not a snake).
While the LLM spits out each ambiguous context search result, it never answers the actual query without a human cheaters help. =3
How do you know the fish is playing? Is he happy, enjoying it? =3
Joel_Mckay 10 hours ago [-]
I agree, part of the ambiguity is also unfairly projecting our own subjective experience onto hapless creatures. =3
VaradD09 12 hours ago [-]
I believe it depends on the LLM itself. Like what model as each model has diff weights and diff data trained onn
dgellow 12 hours ago [-]
I would recommend to read the article, it’s actually more nuanced than the title
Tsarp 14 hours ago [-]
Kinda weird to generalize "LLM". Every lab, every model is different. Has its own biases, reward functions etc.
Centigonal 13 hours ago [-]
[dead]
Founderarcstone 13 hours ago [-]
Great point this will be interesting how this develops.
peter_d_sherman 10 hours ago [-]
Observation: A panel of Judges (multiple Judges), whether it's multiple AI's or not, is fundamentally -- a Jury!
sgt101 11 hours ago [-]
3 tasks?
ex1fm3ta 12 hours ago [-]
I kinda find it funny when I use the advisor on claude code and it agrees with the ideas that the previous model did.
For info: the advisor(s) available are higher end models. For example: you use sonnet, the available advisors are opus and fable. If you use Haiku, the advisor are sonnet, opus and fable.
nekusar 11 hours ago [-]
Betteridges law of headlines says "NO"
troupo 13 hours ago [-]
Without reading the article (doesn't matter if it's pro or contra): no, of course not.
It shouldn't even be a debatable question.
dgellow 12 hours ago [-]
I think you should have read the article first, at minimum the subheader
> Discounting the opinions of LLM judges with highly correlated outputs ensures that panels of judges reflect a true diversity of perspectives.
pessimizer 11 hours ago [-]
This doesn't even make logical sense. Actual judges have highly correlated outputs. This would literally be actively sculpting the range of opinions you want to see.
It's like the idea of political districting that thinks that the aim should be to balance each district between "the two" political parties. You're not doing anything but institutionalizing two political parties and constant conflict. You're setting the range of acceptable opinions, then choosing at random between them. Even more relevantly: when both institutionalized parties have the same opinion, it's considered the correct opinion no matter how much or how little public support it has.
RugnirViking 9 hours ago [-]
> the aim should be to balance each district between "the two" political parties.
I've never heard anyone explicitly advocating for gerrymandering in favor of conflict/variance before? Is that a real thing?
troupo 11 hours ago [-]
The quoted sentence still leads to the same answer: no.
Because there's no "discounting of opinions". They are running a separate LLM to "score" opinions. And the result is still "no" regardless of "lineages" or "sources".
And the end of the article leads me to believe that the entire article and approach is LLM-induced garbage:
--- start quote ---
<Following a list of LLM-like suggestions>
When LLM judges agree, we should ask why. Sometimes agreement is independent evidence. Sometimes it is a shared blind spot. A good aggregation method should be able to tell the difference.
--- end quote ---
novaapi 13 hours ago [-]
[dead]
audelesauvage 10 hours ago [-]
[dead]
asamoahf 14 hours ago [-]
[flagged]
Forgeties79 13 hours ago [-]
the LLM-speak is unbearable
vips7L 12 hours ago [-]
It’s awful. A ton of words to say absolutely nothing.
Forgeties79 8 hours ago [-]
I do not understand why anyone would want an LLM to cosplay as them on a forum. I get maybe (selfish) idle curiosity but beyond that…?
vips7L 5 hours ago [-]
I don’t know why anyone just blatantly copies and paste LLM output. Laziness? Anxiety about being wrong? I don’t know.
You solved one of the largest problems with current LLMs. How is it possible that nobody tried that before?
Because they do. There are already LLMs checking outputs of other LLMs, the bullshit answers that you see are the results of failures on that checks. If you remove all checks LLMs will create hallucinations even more often.
Folks should sue in a class-action lawsuit, any legal firm worth their beautiful walnut desks would seriously be happy take on that constitutionally backed mission. =3
"Nonsensical?" They commit violent crimes, they get prosecuted for said violent crimes, and are serving prison sentences for those crimes. The algorithm picks up on this trend using the same logic that insurance actuaries use, which has also been largely neutered by critical theory.
What even is the argument here-- they're all innocent? Cops are ignoring piles of dead white people and their white murderers to only go patrol brown neighborhoods? We both know neither claim is true. The usual complaint is that cops avoid their neighborhoods and/or are lazy in investigating the crimes they report. The idea of overpolicing has always been a Marxist double-bind...nonsensical, I daresay.
I'm mostly talking about random coding errors.
Drawing a line red to split up the image then has them answer correctly.
Their failure modes are highly correlated.
Citation needed
Try it yourself. Get one to hallucinate, then paste that text into a new window and ask it to verify the facts.
EDIT:
Also - Cohen, Hamri, Geva & Globerson, "LM vs LM: Detecting Factual Errors via Cross Examination": Cross-examination "detects over 70% of the incorrect claims while maintaining a high precision of >80%".
So 70% for ANY error, not just hallucinations.
A fish can play with as many strings as it finds, but only one when on a hook. Yet this too is an incorrect answer, as it again ignores the ambiguity in the phrasing. =3
If a LLM based chat bot does ever answer it correctly, than you know with a fair degree of certainty it was content moderators stepping into the chat. Have a wonderful day. =3
https://en.wikisource.org/wiki/The_Poems_of_John_Godfrey_Sax...
https://en.wikipedia.org/wiki/Blind_men_and_an_elephant
While the LLM spits out each ambiguous context search result, it never answers the actual query without a human cheaters help. =3
https://en.wikipedia.org/wiki/Pareidolia
For info: the advisor(s) available are higher end models. For example: you use sonnet, the available advisors are opus and fable. If you use Haiku, the advisor are sonnet, opus and fable.
It shouldn't even be a debatable question.
> Discounting the opinions of LLM judges with highly correlated outputs ensures that panels of judges reflect a true diversity of perspectives.
It's like the idea of political districting that thinks that the aim should be to balance each district between "the two" political parties. You're not doing anything but institutionalizing two political parties and constant conflict. You're setting the range of acceptable opinions, then choosing at random between them. Even more relevantly: when both institutionalized parties have the same opinion, it's considered the correct opinion no matter how much or how little public support it has.
I've never heard anyone explicitly advocating for gerrymandering in favor of conflict/variance before? Is that a real thing?
Because there's no "discounting of opinions". They are running a separate LLM to "score" opinions. And the result is still "no" regardless of "lineages" or "sources".
And the end of the article leads me to believe that the entire article and approach is LLM-induced garbage:
--- start quote ---
<Following a list of LLM-like suggestions>
When LLM judges agree, we should ask why. Sometimes agreement is independent evidence. Sometimes it is a shared blind spot. A good aggregation method should be able to tell the difference.
--- end quote ---