davidmanheim

6 ms·

1.
by davidmanheim 2mo ago
Basic microeconomic theory can tell us a lot about the questions of where profits will go. Here's an interactive explainer and LLM-friendly source package showing how those dynamics work, for people to play with.
2.
The Economics of LLM Profits: HyperScalers, or Labs?(davidmanheim.com)
2 points by davidmanheim 2mo ago|1 comments
3.
by davidmanheim 1y ago
Yes - I'm fairly culturally conservative, but very much don't support the (so-called) conservative political leaders basically anywhere.
4.
Building less-flawed metrics: Understanding and creating better systems(cell.com)
3 points by davidmanheim 3y ago|0 comments
5.
by davidmanheim 3y ago
Here are some good lists; Rob Wiblin https://medium.com/@robertwiblin/things-i-recommend-you-buy-... Rosie Campbell https://www.facebook.com/ardenlk/posts/10156553178262333 Sam Bowman https:
6.
by davidmanheim 3y ago
...but progress was slower. Exponential growth means it grows faster now than it did then. And yes, relative poverty matters greatly, but absolute poverty is a real thing even in the US. In 1964, the US launched the war on poverty, and pove
7.
by davidmanheim 3y ago
I don't know enough about heat pumps to comment about the substance of this report, but looking at the list of their other reports at the end,it seems worth pointing out the heavily ideological, global warming denying biases of the gro
8.
by davidmanheim 3y ago
It's more specific than that once they actually operationalize the question: This question resolves as YES if, in Meta's 2024 Q4 Quarterly Adversarial Threat Report, Meta claims that there was at least one "coordinated inauth
9.
by davidmanheim 3y ago
> Nowhere do they define "AGI" Ummm, maybe you should have looked? At the top of the very first prediction, here: https://www.metaculus.com/questions/5121/date-of-artificial-... We will thus define &q
10.
by davidmanheim 3y ago
Well, I do know something about those topucs, but here's a twitter thread from someone else who thinks the people worried about risks are silly, who's definitely an expert, with speculation on how he thinks GPT-4 could be made int
11.
by davidmanheim 4y ago
Yeah, and the idea that these programs can't have affordances is silly reliance on a definition that ignores what can happen. It clearly proves too much - humans have a limited action space - they can only move muscles. So they can&#x2
12.
by davidmanheim 4y ago
"The universe is not deterministic." Maybe not. I think that's a big claim, and it's being made, not proved. But let's say it's right, ignoring the massive burden of proof. We can hook up a quantum random numbe
13.
by davidmanheim 4y ago
You sure know a lot about what a set of poorly defined future technologies will and will not do!
14.
by davidmanheim 4y ago
...it sounds like your dog is ruining your productivity. Are you doing anything to adjust your pet's schedule or otherwise deal with that?
15.
by davidmanheim 4y ago
We should probably ask someone who took a heat transfer or other undergrad physics class, but from my basic understanding of the physics, "concentrating heat" isn't really a thing - at the very least, you'd need to extra
16.
by davidmanheim 4y ago
I've spoken with at least half the people they thanked for input, and several of the people cited.
17.
by davidmanheim 4y ago
Ethics is about decisions, and most of the discussions about the "difficulty" of infinite ethics only work if you ignore that. (And so it's particularly frustrating that they didn't bother addressing our work pointing ou
18.
by davidmanheim 4y ago
Yes. Turns out that commercial transactions happen in a social context, and even when it doesn't come back to bite you financially, sometimes paying less has social costs that are far higher than what you "saved".
19.
by davidmanheim 4y ago
Agreed - but it's a larger project than this post. That said, I'd be very interested in people building models of this that are operationalizable and falsifiable! If anyone is interested, they should be in touch.
20.
by davidmanheim 4y ago
Also, why is any pizza ever getting cut into 12 slices?
21.
by davidmanheim 4y ago
It certainly matters that Mesopotamian agriculture had a collapse in ca. 2,200 BCE, but it doesn't matter whether you point out that it was the Akkadian empire founded by Sargon which collapsed, as the OP wanted to correct me about, or
22.
by davidmanheim 4y ago
Thanks for linking, and it's definitely a more polished version of what I was thinking - but after I discussed it with various researchers, as I noted in the post, the forum post is mostly material removed from the paper as interesting
23.
by davidmanheim 4y ago
Thanks for trying to provide feedback, but I think all you're doing is harping on non-universal definitions and noting debated historical points, most of which I'm familiar with but didn't think were worth extended discussion
24.
by davidmanheim 4y ago
No - those systems simulate "control" at the level of giving commands, not actual motor control. So practically, current AI can win at Go and play at superhuman levels, but still cannot drive a car as well as most adults.
25.
by davidmanheim 4y ago
People may want to see the appendix to our RAND report on areas related to DARPA's focus, for more depth on the relevant uses of automation, ML, and AI - https://www.rand.org/pubs/research_reports/RR2489.html
26.
by davidmanheim 5y ago
Is the plastic film this uses actually safe for food?
27.
by davidmanheim 5y ago
Because nothing-at-stake attacks only allow double spending, not arbitrary forged transactions.
28.
by davidmanheim 5y ago
Others pointed out that it's hard to measure subjective factors well. I'd point to a different issue - if you have a metric for, say, wellbeing which is correlated with what you actually care about, putting pressure on other parts
29.
by davidmanheim 5y ago
For those that don't want to click through to the PDF, but want to be able to pretend they read the paper to comment, here's the abstract and conclusion sections: Abstract: Metrics are useful for measuring systems and motivating b
30.
by davidmanheim 5y ago
Not even comments that ignore who the OP is!

More ›