this post was submitted on 15 May 2024
1 points (100.0% liked)

TechTakes

1403 readers
26 users here now

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

founded 1 year ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
[–] [email protected] 0 points 6 months ago (148 children)

From Re-evaluating GPT-4’s bar exam performance (linked in the article):

First, although GPT-4’s UBE score nears the 90th percentile when examining approximate conversions from February administrations of the Illinois Bar Exam, these estimates are heavily skewed towards repeat test-takers who failed the July administration and score significantly lower than the general test-taking population.

Ohhh, that is sneaky!

[–] [email protected] 0 points 6 months ago (145 children)

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!

[–] [email protected] 0 points 6 months ago (56 children)

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.

[–] [email protected] 0 points 6 months ago (2 children)

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.

[–] [email protected] 0 points 6 months ago

Chat GPT had the book entirely memorized, it should have scored perfect almost all the time.

The types of multiple choice questions aren't simple recall of learned facts. It requires application of abstract concepts to new facts, with a lot of red herrings. 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.

Studying for the bar exam starts with memorizing a bunch of rules, but actually getting out and applying them is a separate skill.

[–] [email protected] 0 points 6 months ago (2 children)

Chat GPT had the book entirely memorized

I feel like this exposes a fundamental misunderstanding of how LLMs are trained.

[–] [email protected] 0 points 6 months ago

Don't worry friend, you are correct.

[–] [email protected] 0 points 6 months ago (2 children)

They're auto complete machines. All they fundamentally do is match words together. If it was trained on the answers and still couldn't reproduce the correct word matches, it failed.

[–] [email protected] 0 points 6 months ago (2 children)

You have the energy to spread misinformation and spam downvotes, how about an intelligent responses instead?

[–] [email protected] 0 points 6 months ago

how about fuck off

[–] [email protected] 0 points 6 months ago (1 children)

How about I ban you for being obnoxious instead?

[–] [email protected] 0 points 6 months ago (1 children)

I like them rules you got here

[–] [email protected] 0 points 6 months ago (1 children)

"don't come into our loungeroom and piss on the floor" should be simple and obvious, and yet

[–] [email protected] 0 points 6 months ago

@dgerard @TachyonTele that rug really tied the room together, man

[–] [email protected] 0 points 6 months ago (1 children)

They aren't auto complete machines, they are neural networks. Why are you trying to explain it when you clearly don't have the first idea of how thigs work?

[–] [email protected] 0 points 6 months ago

the very funny thing is, all of the garden variety free text autocomplete systems I’ve worked with have been implemented using neural nets. it’s not like it’s a particularly new or novel approach. but surely the AI bros coming into this thread know that and they’re not just regurgitating buzzwords, right?

load more comments (53 replies)
load more comments (141 replies)
load more comments (143 replies)