There is evidence that high-frequency, 50% accurate bot traders make most of the money on prediction markets simply due to being able to make bets faster.
The way I put this to myself is that AI gives “correct correct answers and incorrect correct answers”.
They almost always generate logically correct text, but sometimes that text has a set of incorrect implicit assumptions and decisions that may not be valid for the use case.
Generating a correct correct solution requires proper definition of the problem, which is arguably more challenging than creating the solution.
> which is arguably more challenging than creating the solution.
This hasn't been the case in my experience. Devising a correct solution without a definition of the problem is impossible because you wouldn't recognize a correct solution without a definition. Often you discover the problem definition by exploratory programming and trial and error on solutions, but LLMs are still good for process this too. Arguably better because they type faster so you can iterate faster!
It’s simpler than that - it’s a guessing machine that has superior access to a whole load of information and capacity to process at a speed at which we humans cannot compete.
Does it make it better than us? No because ultimately the thing itself doesn’t ‘know’ right from wrong.
Yeah, very often the issue is that some context is missing. It'll say something true, but which misses the bigger point, or leads to a suboptimal result. Or it interprets an ambiguous thing in one specific way, when the other meaning makes more sense. You have to keep your wits about you to catch these things.
It's an incredible tool but it's also very derpy sometimes, full of biases, blind spots etc.
Departments base grad school admissions on grant awards. The article states: grant awards for MIT went down more than 20%, then new MIT grad students went down 20%. The decrease in students has nothing to do with academia being detached from industry.
It is a real shame too, because industry is completely incapable of doing basic research. Universities make the fuzzy ideas, and companies turn them into widgets. The only exceptions in history to this are the monopolies, which have their own obvious problems. They cannot produce non-rival, non-excludable goods - stuff that's hard to patent.
Sometimes. I've seen researchers who just churn out useless junk for citation mining and I don't see a lot of overlap between their work and what industry does. That's probably one of the most demoralizing things about academia in my opinion. You sometimes have to be obsequious to people whose goal is just citation farming and whose papers are useless junk filled with buzzwords. I see this a lot in systems and security research. But I also know some researchers who do amazing work and whose research directly gets used in industry.
Yes, I hear you on how academia chases metrics. I would argue this phenomena is not worse than Company Z making a boilerplate AI chat tool that is no more useful than the flagship popular products. I think the fairest comparison is comparing the best researchers in academia/industry. I think they accomplish different things because they have different goals/incentives.
Vaswani, A., et al. (2017) Attention Is All You Need. Proceedings of the 31st International Conference on Neural Information Processing Systems, Long Beach, 4-9 December 2017, 6000-6010.
Basic research would be something like optimal control theory, which came well before the transformer design.
I'm not trying to be evasive; I can see how my distinction could be seen as conveniently just outside industry's purview. Put it this way: I think companies, particularly small ones, are incentivized to pursue well-known methods/materials. Innovation modulates and optimizes.
Obviously there is a selection effect that confounds any causal comparisons between those who do and do not get into MIT. But the better counterfactual is students who are accepted but do not attend. A diff-in-diff study with these two groups would be a better test. There are unique features of MIT: more money, elite network, etc. I do share your skepticism though - I've worked w/ MIT people before. I think they are very smart but also very lucky.
Albert Hirschman wrote a great book about the rhetoric people use to stifle policy proposals 35 years ago. “It’s futile; it won’t ever work” is one common argument. It’s not a meme so much as a cynical reflexive intuition
reply