In my memory, the main impetus was a couple of leading AI safety ML researchers started making the case for 5-year timelines. They were broadly qualitatively correct and remarkably insightful (promoting the scaling-first worldview), but obviously quantitatively too aggressive. And AlphaGo and AlphaZero had freaked people out, too.
A lot of other people at the time (including close advisers to OP folks) had 10-20yr timelines. My subjective impression was that people in the OP orbit generally had more aggressive timelines than Ajeya’s report did.
William_MacAskill
Karma: 10,892
My favourite version of an international AGI project
Forethought series: an international AGI project
Are Short AI Timelines Really Higher-Leverage?
Against Maxipok: existential risk isn’t everything
What sort of post-superintelligence society should we aim for?
Re “Oxford EAs”—Toby Ord is presumably a paradigm of that. In the Great AI Timelines Scare of 2017, I spent some time looking into timelines. His median, then, was 15 years, which has held up pretty well. (And his x-risk probability, as stated in the Precipice, was 10%.)
I think I was wrong in my views on timelines then. But people shouldn’t assume I’m a stand-in for the views of “Oxford EAs”.
I agree—this is a great point. Thanks, Simon!
You are right that the magnitude of rerun risk from alignment should be lower than the probability of misaligned AI doom. However, in worlds in which AI takeover is very likely but that we can’t change that, or in worlds where it’s very unlikely and we can’t change that, those aren’t the interesting worlds, from the perspective of taking action. (Owen and Fin have a post on this topic that should be coming out fairly soon). So, if we’re taking this consideration into account, this should also discount the value of word to reduce misalignment risk today, too.
(Another upshot: bio-risk seems more like chance than uncertainty, so biorisk becomes comparatively more important than you’d think before this consideration.)
Okay, looking at the spectrum again, it still seems to me like I’ve labelled them correctly? Maybe I’m missing something. It’s optimistic if we can retain a knowledge of how to align AGI because then we can just use that knowledge later and we don’t face the same magnitude of risk of the misaligned AI.
I agree with this. One way of seeing that is how many doublings of energy consumption civilisation can have before it needs to move beyond the solar system? The answer to that is about 40 doublings. Which, depending on your views on just how fast explosive industrial expansion goes, could be a pretty long time, e.g. decades.
I do think that non-existential level catastrophes are a big deal even despite the rerun risk consideration, because I expect the civilisation that comes back from such a catastrophe to be on a worse values trajectory than the one we have today. In particular, because the world today is unusually democratic and liberal, and I expect a re-roll of history to result in less democracy than we have today at the current technological level. However, other people have pushed me on that, and I don’t feel like the case here is very strong. There are also obvious reasons why one might be biassed towards having that view.
In contrast, the problem of having to rerun the time of perils is very crisp. It doesn’t seem to me like a disputable upshot at the moment, which puts it in a different category of consideration at least — one that everybody should be on board with.
I’m also genuinely unsure whether non-existential level catastrophe increases or decreases the chance of future existential level catastrophes. One argument that people have made that I don’t put that much stock in is that future generations after the catastrophe would remember it and therefore be more likely to take action to reduce future catastrophes. I don’t find that compelling because I don’t think that the Spanish flu made us more prepared against Covid-19, for example. Let alone that the plagues of Justinian prepared us against Covid-19. However, I’m not seeing other strong arguments in this vein, either.
One more clarification to the comment for forum users: I have tendonitis, and so I’m voice dictating all of my comments, so they might read oddly!
Thanks, that’s a good catch. Really, in the simple model the relevant point of time for the first run should be when the alignment challenge has been solved, even for superintelligence. But that’s before ’reasonably good global governance”.
Of course, there’s an issue that this is trying to model alignment as a binary thing for simplicity, even though really if a catastrophe came when half of the alignment challenge had been solved, that would still be a really big deal for similar reasons to the paper.One additional comment is that this sort of “concepts moving around issue” is one of the things that I’ve found most annoying from AI, and where it happens quite a lot. You need to try and uproot these issues from the text, and this was a case of me missing it.
I think not at all a dumb idea, and I talk about this in Section 6.4. It actually feels like an activity you could do at very low cost that might have very high value per unit cost.
The second way in which this post is an experiment is that it’s an example of what I’ve been calling AI-enhanced writing. The experiment here is to see how much more productive I can be in the research and writing process by relying very heavily on AI assistance — Ttrying to use AI rather than myself wherever I can possibly do so. In this case, I went from having the basic idea to having this draft in about a day of work.
I’d be very interested in people’s comments on how apparent it is that AI was used so extensively in drafting this piece — in particular if there are examples of AI slop that you can find in the text and that I missed.
The first way in which this post is an experiment is that it’s work-in-progress that I’m presenting at a Forethought Research progress meeting. The experiment is just to publish it as a draft and then have the comments that I would normally receive as GoogleDoc comments on this forum post instead. The hope is that by doing this more people can get up to speed with Forethought research earlier than they would have and we can also get more feedback and thoughts at an earlier stage from a wider diversity of people.
I’d welcome takes from Forumites on how valuable or not this was.
Rerunning the Time of Perils
I, of course, agree!
One additional point, as I’m sure you know, is that potentially you can also affect P(things go really well | AI takeover). And actions to increase ΔP(things go really well | AI takeover) might be quite similar to actions that increase ΔP(things go really well | no AI takeover). If so, that’s an additional argument for those actions compared to affecting ΔP(no AI takeover).
Re the formal breakdown, people sometimes miss the BF supplement here which goes into this in a bit more depth. And here’s an excerpt from a forthcoming paper, “Beyond Existential Risk”, in the context of more precisely defining the “Maxipok” principle. What it gives is very similar to your breakdown, and you might find some of the terms in here useful (apologies that some of the formatting is messed up):
”An action x’s overall impact (ΔEVx) is its increase in expected value relative to baseline. We’ll let C refer to the state of existential catastrophe, and b refer to the baseline action. We’ll define, for any action x: Px=P[¬C | x] and Kx=E[V |¬C, x]. We can then break overall impact down as follows:ΔEVx = (Px – Pb) Kb+ Px(Kx– Kb)
We call (Px – Pb) Kb the action’s existential impact and Px(Kx– Kb) the action’s trajectory impact. An action’s existential impact is the portion of its expected value (relative to baseline) that comes from changing the probability of existential catastrophe; an action’s trajectory impact is the portion of its expected value that comes from changing the value of the world conditional on no existential catastrophe occurring.
We can illustrate this graphically, where the areas in the graph represent overall expected value, relative to a scenario with a guarantee of catastrophe:
With these in hand, we can then define:Maxipok (precisified): In the decision situations that are highest-stakes with respect to the longterm future, if an action is near‑best on overall impact, then it is close-to-near‑best on existential impact.
[1] Here’s the derivation. Given the law of total expectation:E[V|x] = P(¬C | x)E[V |¬C, x] + P(C | x)E[V |C, x]
To simplify things (in a way that doesn’t affect our overall argument, and bearing in mind that the “0” is arbitrary), we assume that E[V |C, x] = 0, for all x, so:
E[V|x] = P(¬C | x)E[V |¬C, x]
And, by our definition of the terms:
P(¬C | x)E[V |¬C, x] = PxKx
So:
ΔEVx= E[V|x] – E[V|b] = PxKx – PbKb
Then adding (PxKb – PxKb) to this and rearranging gives us:
ΔEVx = (Px–Pb)Kb + Px(Kx–Kb)”
Thanks! I agree strongly with that.
Merry Christmas, everyone!
This year, I’m feeling grateful to be me.
Recently, I gave some information about myself to Claude, and asked how I compared to other 38-yr-old men in the world.
I thought I understood global inequality well, but I still found the results quite moving. The usual AI sycophancy and reassurance was gone:
I’d appreciated my privilege in income, but hadn’t thought as much about the nature of my work, my health, or my leisure time. I recommend you try it, too (I’ve put a prompt below).
You can also try Giving What We Can’s new Birth Lottery tool — find out what your life would be like if you were born as a random person in the world. When I tried it, I was born in India. On average my life would be around 9 years shorter, with 13 years of schooling instead of 18, and income around 10× lower—even after adjusting for local prices.
I asked Claude to give me a day in the life of a typical 38-year old Indian man:
If you’re feeling privileged this year, consider making a donation to an effective charity—we give gifts to our friends and family at Christmas, so why not give a gift to the world, too.
I’m doing a matching scheme, with a list of great charities, on Substack here and Twitter here, and pasted below, too. Thanks so much to everyone who’s donated so far—currently GiveDirectly and the EA Animal Welfare Fund are in the lead!
And if you want to turn that giving into a regular commitment, consider taking the 10% Pledge — it’s among the single highest-impact, and most personally fulfilling, choices you can make.
My matching scheme: I’m matching donations up to £100,000 (details below), across 10 charities and 6 cause areas. If you want to join, say how much you’re donating and where, as a reply! I’ll run this up until 31st December.
Details of the match:
I’ll give this money whatever happens, so this isn’t increasing the total amount I’m giving to charity.
However, your donations will change *where* I’m giving. I’ll allocate my donations in proportion to the ratio of donations from others as part of the match, with two bits of nuance: 1. I’ll cap donations at £40,000 to any one cause area 2. To prevent extreme ratios, I’ll treat every charity on the list as already having received £1000. The aim of this match is to encourage public giving and public discussion around giving, so I’ll only match people who publicly state that they are giving on here or other social media, as a reply or quote.
The charities:
GiveDirectly: Sends cash directly to people living in extreme poverty, letting recipients decide how best to use it.
Global Health and Development EA Fund: Makes grants to the most effective opportunities in global health and poverty alleviation.
SecureBio: Develops technologies and policies to delay, detect, and defend against pandemics.
The Humane League: Runs corporate campaigns to improve conditions for farmed animals and reduce animal suffering in the food system.
Animal Welfare EA Fund: Makes grants to the most effective opportunities to reduce animal suffering.
METR: Evaluates frontier AI systems for dangerous autonomous capabilities before deployment.
Longterm Future EA Fund: Makes grants to reduce existential risk, with a particular focus on AI safety.
Forethought (note: my own org!): Researches how best to navigate the transition to superintelligent AI.
Eleos AI: Researches AI sentience and welfare, preparing for the possibility that future AI systems may be moral patients.
EA Infrastructure fund: Makes grants to build the effective altruism community and support the infrastructure that helps it function.
And if you want to try my experiment for yourself, here’s a prompt (put the answers in after the questions):