There’s been a massive wave of AIbiosec programs recently. I uncertain how fast this can be scaled up, then again, this isn’t my area.
Chris Leong
Karma: 7,746
I think I might have previously posted this as a comment on another post, but I’ll share here as well:
”Recently I’ve been thinking about how the EA community is not what it used to be and how it’s unlikely to ever be like that again (EA resonated strongly in a social context that no longer exists). But, of course, there’s a saying ‘when one door closes, another opens’. This leads to the question: ’What could EA be instead?’.
I’ll quote from Communities are Not Fungible:
“But a community that took twenty years to develop its particular structure of norms and mutual knowledge cannot be regrown in twenty years, because the conditions that shaped it no longer exist. The people are older, the context has changed, and the specific convergance of circumstances that brought those particular individuals together in that particular configuration at that particular time is gone. Communities are path-dependent in the strongest possible sense: their current state is a function of their entire history, and you can’t rerun the history.”
Contexts change and what resonates during one time period is different from what resonates in another.”
I can see two paths forward:
• Defining a new vision for where we find ourselves today
• Managed decline[1]- ^
Which sounds bad, but sometimes may actually be the best strategy.
- ^
How was the 40% figure calculated?
The funds we’re donating to are aware of AI’s impact on the world and seem likely to take steps to use AI to improve their outcomes
There is definitely some correlation between current competence and ability to adapt to a changing world, though I suspect that there’s also a huge amount of stickiness that undermines this.
All that said, if you think there should be a higher discount, you can apply it, but that doesn’t substantially change the results regarding what goes to AW. Indeed, if you increase the discount to 90%, you don’t get much more to AI (and less overall to GCRs as a whole, likely because the discount also applies to biorisk and nuclear weapons)
Fascinating. Is biorisk depreciating because some of those projects have a long payoff time?
My intuition is that biorisk talent will be more easily redeployable in a world undergoing an AI transition than a lot of the animal talent (more sticky).
I haven’t dived into the figures/calculation, but I’m pretty skeptical about donating 46% to regular animal welfare projects given AI progress. Is the AI discount just based on x-risk or does it account for the possibility of AI radically transforming the world in such a way that the impact of current charities becomes moot?
Been thinking about morality recently. Here are my current thoughts, take them with a grain of salt because they aren’t battle-tested yet.
There are some strong arguments for utilitarianism, but regardless of what is correct theoretically, in practise utilitarianism doesn’t work well without some kind of deontological bars.
Continuing with attempting to develop a pragmatic morality, it then become clear that virtue ethics is important too because a) rules are rigid compared to judgement b) decisions aren’t independent but also affect how you’ll act in the future[1].
Some folks may be quite tepid in integrating virtue ethics, but my intuition is that the more common fault will be to give yourself too much latitude, so you’ll probably want to revive some of your old deontological bars.
I view the next stage after this as introducing a sort of meta-virtue ethics to balance the three components (utilitarianism, deontology and virtue ethics; obviously it would be possible to break this down further). But this likely gives you too much latitude again, so you’ll probably want to introduce some kind of meta-deontology to limit how you update the balance.
You could go further than this, but you’d probably be running into decreasing marginal utility.
Is EA Significantly Underestimating the Amount of Suffering in the World?
I assume this is going forward and there’s no requirement to backlabel?
AI Risk Agility Plans—v0.1
One thing that’s unclear to me is whether attempts to use AI systems to augment human capabilities in these domains is in-scope or whether the round is focused on direct enhancement of these capabilities.
I’m also curious about your opinion on whether biological-enhancement based approaches are likely to bear fruit in time to matter. Do you think it’s plausible that timelines might be long on our current path or are you more hoping that there’s a pause that provides humanity with more time?
(Alternatively, is it more that you think that we need enhanced capabilities to succeed at alignment even if current timeline projections makes this appear challenging?).
I suspect that the main use of forecasting is if you need a probability for something and you don’t really have time to look into it yourself or you wouldn’t trust your judgement even if you did.
I probably should have clarified that EA as a community building effort isn’t drawing the same talent. That said, talent that was attracted through community building efforts earlier has had more time to “level up” (as you mentioned) and orgs have likely improved their ability to recruit experienced professionals directly compared to the past (though my intuition is that some orgs haven’t fully appreciated the costs of weakening value-alignment since these impacts take a long time to emerge).
A further caveat: I don’t know exactly how things are at elite universities these days.
For fairness, I’ll just add a comment that the following edits were made after the competition deadline:
• “The first two are based on this post”
• “If doing what worked in other cities worked, it would have already worked”
• “—but it synergises nonetheless”I made some additional edits after it was announced this essay came in second for this question.
I’ll PM you and post my comment publically later. I’m curious if anyone else makes similar points if I don’t do so.
“Every entity in the ecosystem gets a living profile on the platform”—Given that this has a high probability of having have ecosystem wide impacts, how much analysis have you conducted of the downsides of doing this? Are you willing to share this analysis?
Building the EA/AI Safety Scene in San Francisco
I notice myself feeling negative towards these laws.
This isn’t even an absolute rejection on my part, I’m just worried that governments are sliding too much towards paternalism without properly considering the downsides of their legislation.
Sounds, like it’s a recommendation, not a rule?
Do you want paragraphs or are nested dot points fine?
Btw, I just thought I should say that I really appreciate you folks writing this post. I don’t want you to think that I disliked your article just because I disagreed on one point (which is how things can sometimes come off).
Using the term “cult” in in a purported terminology post isn’t helpful.
I know it’s annoying, but I would have suggested splitting this into two posts, one introducing the terminology in a neutral fashion and another where you use the terminology to make your criticism.