Project Idea: 'Cost to save a life' interactive calculator promotion

Some clarification questions (which I hope will help people preparing these applications).

1. For standard grants ($100k-$1m), you say next steps "may include follow-up questions or a direct decision." Roughly how much of the decision do you expect to depend on the EoI text itself? I'd find find it useful to know how much of the case for credibility and theory of change (etc.) needs to fit in 2,000 characters, versus how much can be introduced at a later stage (if one gets there).

2. Relatedly, is there a mechanism to link to supporting material (a longer scope note a more detailed budget, prior work, etc.), or would you rather it all fits in the 2k character limit?

3. Your AI governance RFP said the EoI stage focuses mainly on the first three assessment criteria (ToC, track record, strategic judgment), with the rest covered at full review. Is there an analogous shortlist here (going beyond the ITN framework mentioned)?

Yikes, something is wrong here -- I now put it to 50 and it says 0? I'll ask the forum moderators

Quick follow-up: I've worked with Claude Code (several iterations; wanted Fable, got blocked so pushed to Opus) to put up this page comparing and bridging Pablo's model to ours...

• where and how the two overlap and differ
• a simplified calculator that feeds a cost/kg into a stripped-down version of the cost→share logic.

This is early -- I continue vetting and working on it and would love your feedback

There's a potential complementarity: we focus on is the cost distribution, Pablo on the demand response is Pablo's

(If you add hypothes.is comments there I'll respond and adapt.)

Thanks for this! We've been doing some related work through The Unjournal Pivotal questions -- see this modeling interface/calculator. I'm reading through your modeling and discussion carefully now, aiming to do some comparisons between our approaches, and it might be worth some conversation/collaboration (feel free to DM).

Also see the resources and evidence coming out of our recent workshop, which we're following up on, aiming to get a wider range of views and a productive crux-mappiung conversation between 'skeptics and optimists'. We plan to share further synthesis of stated beliefs as it comes together.

(Also see the how it works explainer and the TEA comparisons. Much/most of these were AI generated with repeated feedback, and may benefit from further human scrutiny.

Wait back to 50% because it says “end factory farming” rather than “ substantially reduce factory farming”

I voted 25% aka alt proteins more likely. Not sure what happened w the interface

Intuition that moral norms are hard to change with personal interests and tastes and culture being very much in the opposite direction. Yes, we have seen changes in these moral norms towards LGBT, racism etc, but animals can't speak for themselves in the same way

I sometimes felt you were implying "no evidence" when I thought something closer to "very weak evidence" would be more appropriate.

Can you provide some specific examples, and I'll reconsider

However, in practice, I would agree the focus should overwhelmingly be on decreasing the uncertainty about the extent to which AI models have welfare, and how to measure it.

I'm not sure if we 'agree' here, or at least that's not the point I was making. I was accepting that AI models (either the information flow itself, or the physical electron flows or something) might indeed "have" or generate consciousness and welfare.

My main doubt was more:

• Suppose this valenced experience and 'welfare' indeed comes about in such a system, which is so very different from human or other biological systems,
• How could we evenr know, even in an expected value directional sense, what things made this welfare higher or lower
  • or whether it is even 'net positive' so we want to 'make more of these' or the opposite
• We cannot trust the 'talker' to tell us this and I don't understand what other signals coming out of the AI would be reliable evidence
• This would imply we have 'deep uncertainty' about the impact of our actions on anything welfare-relevant, making our choices morally irrelevant.

So I'm not sure that we have useful ways to decrease the uncertainty about whether th AI models have welfare, and I'm not sure we really have any ways to measure it. But I guess agree we should try to decrease the uncertainty about 'whether we can now or will ever have ways to measure it' (or 'it's valence') ... which is what my post was sort of trying to do.

Perhaps the strongest argument against this would be "what if the model says it is conscious, it is in pain, the data suggests it is not lying and that it is confident in its statement?" I don't find this convincing. ... Even if the talker has no access to the valenced consciousness, it's model may simply lead it to a confident and wrong answer about this.

Did you mean "If" instead of "Even if"?

What I meant was that even if the AI is making a confident statement that it is in pain, and it is not deliberately lying, it doesn't mean that it actually knows the answer to 'is it (or is anything in it's system) in pain'.

So the 'even if' was setting off a contrast between 'not having a access to the answer' and 'making a confident statement'.

Thanks. I've read much of that "bullshit" paper and I think there's some interesting overlap. Some ways I think it relates:

Their "lack of external validation of AI welfare" problem seems at the root of the problem I name that "the signals we get from the AI may not tells us anything directional about the valence of the part of the AI that has consciousness" (if it does). If there is no ground truth (opportunity for 'falsification') here I don't see how we can credibly make that link.

The evidence they cite about the sensitivity of the measures signals of valence to seemingly irrelevant engineering choices reinforces my doubts.