Top
Best
New

Posted by porridgeraisin 1 day ago

I struggle to believe that AI R&D is ingenious(www.lesswrong.com)
212 points | 210 commentspage 3
cthalupa 1 day ago|
I agree that it probably isn't ingenious, but this article does nothing to actually prove that point. "It didn't do stuff, the stuff it did was low value, I will provide no actual details about any of it."

This a twitter thread, not an article arguing a point in a salient manner.

amelius 1 day ago||
It only appears to be not ingenious because it does not know what you want to achieve.
muestramuertos 20 hours ago||
this really hit me with Jev. it never occurred to me that something like scoring rules weren't applied before.
jgalt212 1 day ago||
The way I look at AI R&D is it's a really clever brute force search of the problem and solution space. The amazing bit to me, is how rarely, these models and methods seems to get stuck in local maxima. i.e. ex ante, I'd expect these models and methods to get stuck in local maxima 100% of the time.
syngrog66 1 day ago||
I'm old enough to have realized that I've done plenty of R&D in my life that wasn't formally named that. R&D is about the substance, the mode, the impact, not the names or titles.
darepublic 1 day ago||
Silent Bob guy just "rebuts" the original poster by spouting how many loc Claude spits out without bugs and unread tests. Which doesn't truly counter anything op said. I'm sure silent on guy can be smarter than me but come on. And yes ai report writing is often verbose and full of crap. Shouldn't need special prompts for it to be focused get to the point
fedeb95 1 day ago||
They are useful if properly guided. Just as every other tool.
chrisjj 1 day ago||
> I struggle to believe at an intuitive level that autonomous R&D is so ingenious.

Why would anyone go to the effort of struggling?

bitwize 19 hours ago||
The bitter lesson is that ingenuity is not really necessary in this realm; more benefits will accrue to you if you just scale harder.

Ingenuity was last decade's moat. But human ingenuity didn't crack those millennium math problems; throwing more money at compute did.

MisterMunchkin 1 day ago|
LLMs are great at doing work quickly and they don’t say something is impossible or discourage you, so you can really tackle problems that previously would have taken years of work.

They aren’t capable of generating truly innovative ideas though. Ask it to come up with a name, it will come up with the same weird slop names over and over. Always gives people the surname Okafor. Because it works by generating probable answers, it can only generate probable names.

nfw2 1 day ago||
I think the reason models' default creative output is so bland is sort of like why if you combine all color crayons it just turn out brown.

These models actually can create content of myriad styles very well. But artistic styles are mutually incompatible. It can't create something that matches all of them, and it doesn't have an opinion on which is best, so bland requests just give bland results. To get good results you need to direct the model into the region of its latent space that makes stuff you thing is good.

Eueudhsbsj32 1 day ago||
> Because it works by generating probable answers, it can only generate probable names.

That's not true anymore. I've asked it to generate names for some LLCs and it came up with plenty of novel names that sounded vaguely Latin and had a certain sound that I had asked for.

Personability 1 day ago||
Could you give some examples? Because I doubt that they sound properly novel.
More comments...