Do you have any data on whether outcomes are improving over time? For example, % published / employed / etc 12 months after a given batch
gw
Karma: 150
I agree! This is mostly focused on the “getting a job” part though, which typically doesn’t end up testing those other things you mention. I think this is the thing I’m gesturing at when I say that there are valid reasons to think that the software interview process feels like it’s missing important details.
So you want to work on technical AI safety
This might look like building influence / a career in the federal orgs that would be involved in nationalization, rather than a startup. Seems like positioning yourself to be in charge of nationalized projects would be the highest impact?
Your GitHub link is broken, it includes the period in the url.
I
Love
Interesting
Alignment
Donferences
I spoke with some people last fall who were planning to do this, perhaps it’s the same people. I think the idea (at least, as stated) was to commercialize regulatory software to fund some alignment work. At the time, they were going by Nomos AI, and it looks like they’ve since renamed to Norm AI.
+ the obvious fact that it might matter to the kid that they’re going to die
(edit: fwiw I broadly think people who want to have kids should have kids)
Hmm, I have exactly one idea. Are you pressing shift+enter to new line? For me, if I do shift+enter
>! I don’t get a spoilerBut if I hit regular enter then type >!, the spoiler tag pops up as I’m typing (don’t need to wait to submit the question for it to appear)
Are you thinking of
Until Dawn?
(also it seems like I can get a spoiler tag to work in comments by starting a line with >! but not by putting text into :::spoiler [text] :::)
Interesting, thanks for the detailed responses here and above!
Here’s a handwavy attempt from another angle:
Suppose you have a container of gas and you can somehow run time at 2x speed in that container. It would be obvious that from an external observer’s point of view (where time is running at 1x speed) that sound would appear to travel 2x as fast from one end of the container to the other. But to the external observer, running time at 2x speed is indistinguishable from doubling the velocity of each gas molecule at 1x speed. So increasing the velocity of molecules (and therefore the temperature) should cause sound to travel faster.
(Also, for more questions like this, see this post on Thinking Physics)
If I make the room bigger or smaller while holding T and P constant, v(sound) does not change. If it did, it would be very obvious in daily life.
This feels a bit too handwavy to me, I could say the same thing about temperature: if the speed of sound were affected by making a room hotter or colder, it would be very obvious in daily life, therefore the speed of sound doesn’t depend on temperature. But it isn’t obvious in daily life that the speed of sound changes based on temperature either.
So now let’s increase T. It doesn’t matter what effect this has on P and V and n, as seen in the above. So what’s left? Increasing T linearly increases the average kinetic energy of the gas molecules (PV and NkT both have units of energy, this is why), and velocity increases as the sqrt of kinetic energy. So if gas molecule velocity is what determines v(sound), then it has to be that v(sound) increases as sqrt(T).
I think this also falls short of justifying that v(sound) increases as T increases. Why does it have to be that v(sound) increases with gas molecule velocity and not decreases instead? Why is it the case that gas molecule velocity determines v(sound) at all?
Worth noting that the scam attempt failed. We keep hearing ‘I almost fell for it’ and keep not hearing from anyone who actually lost money.
Here’s a story where someone lost quite a lot of money through an AI-powered scam:
We can question things, how it went this way or why we are all here with this problem now—but it does not in add anything IMHO.
I think it adds something. It’s a bit strongly worded, but another way to see this is “could we have done any better, and if so, why?” Asking how we could have done better in the past lets us see ways to do better in the future.
This post comes to mind as relevant: Concentration of Force
The effectiveness of force application often depends on its concentration—on whether you can amass locally superior force at the actual decisive moment.
As someone who is definitely not a political expert (and not from or super familiar with the UK), my guess would be that you just can’t muster up enough political capital or will to try again. Taxpayer money (in the US at least) seems highly scrutinized, you typically can’t just fail with a lot of money and have no one say anything about it.
So then if the first try does fail, then it requires more political capital to push for allocating a bunch of money again, and failing again looks really bad for anyone who led or supported that effort. Politicians seem to care about career risk, and all this makes the risk associated with a second shot higher than the first.
I’d agree that this makes a second shot unlikely (including from other governments, if it fails spectacularly enough), if circumstances stay about the same. But circumstances will probably change, so IMO we might eventually get more such taskforces, just not soon.
Is it possible to purchase the 2018 annual review books anywhere? I can find an Amazon link for the 2019 in stock, but the 2018 is out of stock (is that indefinite?).
Re: “up-skilling”: I think this is underestimating the value of developing maturity in an area before trying to do novel research. These are two separate skills, and developing both simultaneously from scratch doesn’t seem like the fastest path to proficiency to me. Difficulties often multiply.
There is a long standing certification for “proving you’ve learned to do novel research”, the PhD. A prospective student would find it difficult to enter a grad program without any relevant coursework, and it’s not because those institutions think they have equal chances of success as a student who does.
Thanks for sharing, I was curious if you could elaborate on this (e.g. if there are examples of AI policy work funded by OP that come to mind that are clearly left of center). I am not familiar with policy, but my one data point is the Horizon Fellowship, which is non-partisan and intentionally places congressional fellows in both Democratic and Republican offices. This straightforwardly seems to me like a case where they are trying to engage with people on the right, though maybe you mean not-right-of-center at the organizational level? In general though, (in my limited exposure) I don’t model any AI governance orgs as having a particular political affiliation (which might just be because I’m uninformed / ignorant).