Agreed. I feel like Alphabet (who wrote this post) has kind of lost track of what they're espousing here.
Self driving cars were the "monkey" 10 years ago but now they are largely derisked. I'd like to see Alphabet working on the next generation of innovations. For me personally, I'd like to see them work on:
- Teleportation.
- Talking animals.
- A cure for death.
I'm not seeing them work on anything remotely as ambitious these days!
> The authors warn the consequences are already becoming visible. AI-generated papers could overwhelm peer-review systems with low-quality work …
It seems like a key problem here is that peer-review is expected but not explicitly funded/rewarded while it is probably one of the aspects where humans still add a lot of value. Academia’s incentives are hugely misaligned (… as usual unfortunately).
Math is one field where you can mechanically prove a paper's findings. The only thing that would need to be judged is the (verified) statement's importance.
Yeah 3 - 4 is typical in STEM at Imperial, depends on the scholarship or funding source. The standard funding tends to assume 3 - 3.5 years, but I vaguely recall that in some departments supervisors had a habit of forcing people to stick around for a few months without funding.
I was in a similar situation as the parent post and skipped/was moved up by two years of high school.
I think it was very beneficial to have to work hard to catch up with more advanced classes. I feel flexibility around this is something parents and schools should take seriously.
(Tbf I was also super lucky to find a very accepting group of nerdy friends in the new year that would tolerate someone younger.)
There’s also a vague argument around hedging some actual risks that some market participants genuinely want to hedge… which depends a lot on the specific bet. Eg hedging exposure to specific political events, wars or even company announcements can be relevant and worth a premium for non-insiders. Where there’s a premium to be collected there are speculators to do so.
> “Hedge funds invest a ton in "alternative data", like credit card transaction data or satellite-imagery (are Walmart's parking lots full?) and need to process as much relevant information as possible to make predictions that are relevant to investments. “
Ah yes the famous credit card data and Walmart parking lots example that hedge funds were giving a few years ago in every interview and news article. Safe to assume that specifically these data sets are not what you should look at to make money.
"High variance, slightly-negative EV shots on a long time horizon" is a gambling addict's way of justifying the old adage, "sure, we're losing money on each sale, but we'll make up for it in volume!"
It's possible but unlikely given the short timeline, diverse questions that require multiple matheamticians, and low stakes. Also they've already run preliminary tests.
> It's possible but unlikely given the short timeline
Yep. "possible but unlikely" was my take too. As another person commented, this isn't really a benchmark, and as long as that's clear, it seems fair. My only fear is that some submissions may be AI-assisted rather than fully AI-generated, with crucial insights coming from experienced mathematicians. That's still a real achievement even if it's human + AI collaboration. But I fear that the nuance would be lost on news media and they'll publish news about the dawn of fully autonomous math reasoning.
Interesting topic and seems like a reasonable thing to study.
It would be cool to see a study where participants behave in a way that leads to more light exposure in month A and the opposite in month B, randomising the time ordering etc.
Despite power analysis and all the “50 people convenience sample, mostly observing based on what people do anyway” seems a bit like it won’t really lead to any action guiding outcomes beyond vaguely confirming people’s priors? As in, perhaps this style of research is not ambitious enough in a way?
reply