Remember how ChatGPT totally aced the bar exam? Wow! yeah, turns out that was just a lie
2 years ago by David Gerard to c/techtakes
archive https://archive.ph/is57b
What I find delightful about this is that I already wasn't impressed! Because, as the paper goes on to say
Moreover, although the UBE is a closed-book exam for humans, GPT-4’s huge training corpus largely distilled in its parameters means that it can effectively take the UBE “open-book”
And here I was thinking it not getting a perfect score on multiple-choice questions was already damning. But apparently it doesn't even get a particularly good score!
[...W]hen examining only those who passed the exam (i.e. licensed or license-pending attorneys), GPT-4’s performance is estimated to drop to 48th percentile overall, and 15th percentile on essays.
officially Not The Worst™, so clearly AI is going to take over law and governments any day now
also. what the hell is going on in that other reply thread. just a parade of people incorrecting each other going "LLM's don't work like [bad analogy], they work like [even worse analogy]". did we hit too many buzzwords?
"Nooo you don't get it, LLMs are supposed to be shit"
But LLM’s don’t work like Typewriters, they work like Microwaves!
I was considering interjecting in there but I don’t want to get it on my clothes, so I’m content just watching from the outside.
Not great, but I’m also not obligated to teach anyone anything, soooooo
Not the worst? 48th percentile is basically "average lawyer". I don't need a Supreme Court lawyer to argue my parking ticket. And if you train the LLM with specific case law and use RAG can get much better.
In a worst case scenario if my local lawyer can use AI to generate a letter and just quickly go through it to make sure it didn't hallucinate, they can process more clients, offer faster service and cheaper prices. Maybe not a revolution but still a win.
48th percentile is basically "average lawyer".
good thing all of law is just answering multiple-choice tests
I don't need a Supreme Court lawyer to argue my parking ticket.
because judges looooove reading AI garbage and will definitely be willing to work with someone who is just repeatedly stuffing legal-sounding keywords into google docs and mashing "generate"
And if you train the LLM with specific case law and use RAG can get much better.
"guys our keyword-stuffing techniques aren't working, we need a system to stuff EVEN MORE KEYWORDS into the keyword reassembler"
In a worst case scenario if my local lawyer can use AI to generate a letter
oh i would love to read those court documents
and just quickly go through it to make sure it didn't hallucinate
wow, negative time saved! okay so your lawyer has to read and parse several paragraphs of statistical word salad, scrap 80+% of it because it's legalese-flavored gobbledygook, and then try to write around and reformat the remaining 20% into something that's syntactically and legally coherent -- you know, the thing their profession is literally on the line for. good idea
what promptfondlers continuously seem to fail to understand is that verification is the hard step. literally anyone on the planet can write a legal letter if they don't care about its quality or the ramifications of sending it to a judge in their criminal defense trial. part of being a lawyer is being able to tell actual legal arguments from bullshit, and when you hire an attorney, that is the skill you are paying for. not how many paragraphs of bullshit they can spit out per minute
they can process more clients, offer faster service and cheaper prices. Maybe not a revolution but still a win.
"but the line is going up!! see?! sure we're constantly losing cases and/or getting them thrown out because we're spamming documents full of nonsense at the court clerk, but we're doing it so quickly!!"
In a worst case scenario if my local lawyer can use AI to generate a letter and just quickly go through it to make sure it didn’t hallucinate
at which point it's just easier to do the right thing straight away, that is pay a lawyer to do their job https://www.bbc.com/news/world-us-canada-65735769
In a worst case scenario if my local lawyer can use AI to generate a letter and just quickly go through it to make sure it didn’t hallucinate, they can process more clients, offer faster service and cheaper prices.
It's a good thing people are so good at vigilance tasks and don't tend to fall onto just relying on the automation.
That’s like saying a person reading a book before a quiz is doing it open book because they have the memory of reading that book.
Except it’s not, because they can’t perfectly recall everything.
It’s more like reading every book in the world, and someone asking you what comes next after “And I…”.
I'm not a big AI guy but it's really not quite like that, models do NOT contain all the data they were trained on.
Edit: I have no idea what's going on down below this comment
I'm not even going to engage in this thread cause it's a tar pit, but I do think I have the appropriate analogy.
When taking certain exams in my CS programme you were allowed to have notes but with two restrictions:
- Have to be handwritten;
- have to fit on a single A4 page.
The idea was that you needed to actually put a lot of work into making it, since the entire material was obviously the size of a fucking book and not an A4 page, and you couldn't just print/copy it from somewhere. So you really needed to distill the information and make a thought map or an index for yourself.
Compare that to an ML model that is allowed to train on data however long it wants, as long as the result is a fixed-dimension matrix with parameters that helps it answer questions with high reliability.
It's not the same as an open book, but it's definitely not closed book either. And the LLMs have billions of parameters in the matrix, literal gigabytes of data on their notes. The entire text of War and Peace is ~3MB for comparison. An LLM is a library of trained notes.
My question to you is how is it different than a human in this regard? I would go to class, study the material, hope to retain it, so I could then apply that knowledge on the test.
The ai is trained on the data, "hopes" to retain it, so it can apply it on the test. It's not storing the book, so what's the actual difference?
And if you have an answer to that, my follow up would be "what's the effective difference?" If we stick an ai and a human in a closed room and give them a test, why does it matter the intricacies of how they are storing and recalling the data?
I'm not even going to engage in this thread cause it's a tar pit, but I do think I have the appropriate analogy.
Proceeds to actively engage in the thread multiple times
The word parameters here must be defined. Is it the weight they are talking about or the input being used to answer the question? For the former, yeah, it's like a person was reading a book and not an open book at all. But if it were used in the input, then it is practically an open book. They have the context on the same input.
Why is that a criticism? This is how it works for humans too: we study, we learn the stuff, and then try to recall it during tests. We've been trained on the data too, for neither a human nor an ai would be able to do well on the test without learning it first.
This is part of what makes ai so "scary" that it can basically know so much.
Well... I do agree with you but human brains are basically big prediction engines that use lookup tables, experience, to navigate around life. Obviously a super simplification, and LLMs are nowhere near humans, but it is quite a step in the direction.
Yeah but neither did Socrates
I guess it comes down to a philosophical question as to what "know" actually means.
But from my perspective is that it certainly knows some things. It knows how to determine what I'm asking, and it clearly knows how to formulate a response by stitching together information. Is it perfect? No. But neither are humans, we mistakenly believe we know things all the time, and miscommunications are quite common.
But this is why I asked the follow up question...what's the effective difference? Don't get me wrong, they clearly have a lot of flaws right now. But my 8 year old had a lot of flaws too, and I assume both will get better with age.
Because a machine that "forgets" stuff it reads seems rather useless... considering it was a multiple choice style exam and, as a machine, Chat GPT had the book entirely memorized, it should have scored perfect almost all the time.
I don't think you understand the type of multiple choice questions involved. Here's a real question:
A father lived with his son, who was an alcoholic. When drunk, the son often became violent and physically abused his father. As a result, the father always lived in fear. One night, the father heard his son on the front stoop making loud obscene remarks. The father was certain that his son was drunk and was terrified that he would be physically beaten again. In his fear, he bolted the front door and took out a revolver. When the son discovered that the door was bolted, he kicked it down. As the son burst through the front door, his father shot him four times in the chest, killing him. In fact, the son was not under the influence of alcohol or any drug and did not intend to harm his father.
At trial, the father presented the above facts and asked the judge to instruct the jury on self-defense.
How should the judge instruct the jury with respect to self-defense?
(A) Give the self-defense instruction, because it expresses the defense’s theory of the case.
(B) Give the self-defense instruction, because the evidence is sufficient to raise the defense.
(C) Deny the self-defense instruction, because the father was not in imminent danger from his son.
(D) Deny the self-defense instruction, because the father used excessive force.
Memorizing the book itself doesn't teach how to answer this type of question. It requires actual application of concepts to the new facts being given.
Yes that is indeed the sort of question I was expecting. But anyway good thing the LLM didn't have just one book, but oodles of books and expert opinion and past exam data at its disposal! Oh wait it didn't help and the machine especially made to give correct answers failed to give correct answers :(
Lol, it literally takes 5s to search for this, come up with the book and read
The defendant has offered evidence of having acted in self-defense.
as the first sentence. How could you claim having that memorised wouldn't help?
90th percentile means it performed equal or better than 90% of the comparisons, no? Not that it got 90% score.
AI being pushed by scam artists...Gee. Who could have guessed?
I did. I guessed. I expressed skepticism when that headline first appeared.
I mean, it’s not shit at everything; it can be quite useful in the right context (GitHub Copilot is a prime example). Still, it doesn’t surprise me that these first-party LLM benchmarks are full of smoke and mirrors.
Not to be confused with Microsoft Copilot, which I have yet to find a use for. Do you not like GH Copilot either?
it’s always fucking “boilerplate” with these assholes, isn’t it? I don’t know how so many people got into this field and didn’t figure out the template, snippet, or macro engines in their editors
That GitHub Copilot and friends are useful? I would argue that their utility is rather subjective, but there are indications that it improves developer productivity.
I’m unsure if you’ve used tools like GH Copilot before, but it primarily operates through “completions” (“spicy autocorrect” in its truest form) rather than a chatbot-like interface. It’s mostly good for filling out boilerplate and code that has a single obvious solution; not game-changing intelligence by any means, but useful in relieving the programmer of various menial tasks.
May I ask, what evidence are you hoping to see in particular?
It's okay at creating writing prompts for me. That's about it.
the perils of hitting /all
dj khaleb suffering from success dot jpeg
I asked AI to summarize the article since it's paywalled. It didn't say anything about lying, should I trust it?
*Politics. Ftfy
sort by controversial did not disappoint :)
this is sooo silicon valley
It's almost like we can't make a machine conscious until we know what makes a human conscious, and it's obvious Emergentism is bullshit because making machines smarter doesn't make them conscious
Time to start listening to Roger Penrose's Orch-OR theory as the evidence piles up - https://pubs.acs.org/doi/10.1021/acs.jpcb.3c07936
The given link contains exactly zero evidence in favor of Orchestrated Objective Reduction — "something interesting observed in vitro using UV spectroscopy" is a far cry from anything having biological relevance, let alone significance for understanding consciousness. And it's not like Orch-OR deserves the lofty label of theory, anyway; it's an ill-defined, under-specified, ad hoc proposal to throw out quantum mechanics and replace it with something else.
The fact that programs built to do spicy autocomplete turn out to do spicy autocomplete has, as far as I can tell, zero implications for any theory of consciousness one way or the other.
Bro the main objection to Orch-OR is that the brain is too warm for Quatnum stuff to happen there, and then they found Quantum stuff in the brain.... So... not sure how it's not suggestive of the reality of Orch-OR
Edit: Btw, I don't know where you're getting the idea that Orch-OR is "Trying to throw out Quantum Mechanics and replace it with something else", considering that it's dependent upon Quantum Mechanics, and we have demonstrated that "Quantum Biology" is a thing in plants - https://www.scientificamerican.com/... and in birds - https://www.nature.com/articles/d41586-021-01725-1
So why not the brain?
it’s very important to me that you don’t type the words “Blake Stacey” into a search engine while explaining quote unquote Quatnum stuff to them
randoms from /all wandering into the vale of sneers: https://www.buttersafe.com/2008/10/23/the-detour/
Oh I see... I didn't realize you were trying to tell me I was talking to Blake Stacey or that he was respected in Quantum Mechanics. I completely misinterpreted what you were trying to tell me. I blame it on the inability of text to properly convey sarcasm.
Kludging an "objective reduction" process into the dynamics is throwing out quantum mechanics and replacing it with something else. And because Orch-OR is not quantum mechanics, every observation that a quantum effect might be biologically important somewhere is irrelevant. Orch-OR isn't "quantum biology", it's pixie-dust biology.
Orch-OR
Never heard of this thing but just reading through the wiki
An essential feature of Penrose's theory is that the choice of states when objective reduction occurs is selected neither randomly (as are choices following wave function collapse) nor algorithmically. Rather, states are selected by a "non-computable" influence embedded in the Planck scale of spacetime geometry.
Neither randomly nor alorithmically, rather magically. Like really, what the fuck else could you mean by "non-computable" in there that would be distinguishable from magic?
Penrose claimed that such information is Platonic, representing pure mathematical truths, which relates to Penrose's ideas concerning the three worlds: the physical, the mental, and the Platonic mathematical world. In Shadows of the Mind (1994), Penrose briefly indicates that this Platonic world could also include aesthetic and ethical values, but he does not commit to this further hypothesis.
And this is just crankery with absolutely no mathematical meaning. Also pure mathematical truths are not outside of the physical world, what the fuck would that even mean bro.
I thought Penrose was a smart physicist, the hell is he doing peddling this.
it's well outside of his ballpark somehow, it's like how Linus Pauling started all that megadose vitamin horseshit (starting with vit C), it sorta, kinda made a vibe-based shred of sense when you ignore all actual details, but he was hopelessly lost because he was not a biologist. what he had was nobel prize so he had enough cred for people to fall for it. many such cases!
I thought Penrose was a smart physicist, the hell is he doing peddling this.
You're right that consciousness and intelligence are not the same. Our language tends to conflate the two.
However, evolution created consciousness over billions of years by emergent factors and no source of specific direction besides being more successful at reproduction. We can likely get there orders of magnitude faster than evolution could. The big problem would be recognizing it for what it is when it's here.
I mean, assuming it is at all possible (or rather that the problem even means anything), I suppose four billion years is a rather generous deadline.
@WolfLink so you're saying there's a measurable correlation between practicing a skill and getting better at it? Amazing
What's this got to do with the Big Averaging Machine?
I just assumed that train on an answer sheet could probably get you past most tests.
@awful.systems
Big brain tech dude got yet another clueless take over at HackerNews etc? Here's the place to vent. Orange site, VC foolishness, all welcome.
This is not debate club. Unless it’s amusing debate.
For actually-good tech, you want our NotAwfulTech community
go to feed...
@awful.systems
Big brain tech dude got yet another clueless take over at HackerNews etc? Here's the place to vent. Orange site, VC foolishness, all welcome.
This is not debate club. Unless it’s amusing debate.
For actually-good tech, you want our NotAwfulTech community
go to feed...
From Re-evaluating GPT-4’s bar exam performance (linked in the article):
Ohhh, that is sneaky!
save