Quick FYI for anyone still waiting to hear back re: EAG London. I learned today that my acceptance email from March never reached me, and apparently there have been a few email hiccups this year.

Might be worth checking spam/alternative inboxes or contacting admissions if you're still waiting. 

What do people think the implications are for the AI safety field if internal deployment is where frontier AI labs are heading? It would seem less tractable to me, especially for people who aren’t already deeply networked into the labs, which I expect describes most people currently trying to pivot into AI safety. 

I'd like to know too. 

Hi! Congratulations on the growth. A few questions from a fairly interested outside observer:

1. How do you think about counterfactual impact? It seems quite difficult to separate outcomes that would likely have happened anyway from outcomes meaningfully attributable to your interventions, particularly in a tough market. I don’t expect that’s fully solvable, but I’m curious about what methods or heuristics you use.

2. What do you see as the main differences between Probably Good and other career advisory organisations/services such as 80,000 Hours or High Impact Professionals? Where do you think your comparative advantage is emerging?

3. You mention an extensive search before selecting an internal hire for the ED role. Roughly how many applicants were involved, and how much staff time did the process consume overall? I’m generally sympathetic to internal hires/closed rounds in most cases because of the potential time/cost savings, but I’d be interested to better understand the level of those potential savings in practice.

Thanks for sharing, I’d definitely be interested in reading a longer retrospective at some point.
 

One angle I’d add is that there may be a complementary, more immediate approach: bringing in experienced grantmakers from adjacent fields and helping them transition into AI safety.

I think many of the core skills (evaluating proposals, making judgment calls with uncertain information, communicating decisions, managing portfolios, etc) are quite transferable. The main gap is context, for the reasons you've described. I think that'd be faster to remedy than building grantmakers from scratch.

Do you think this kind of approach (getting proven grantmakers into AI safety) could help solve the expected bottleneck? Perhaps I’m underestimating the barriers.  

Ok, my updated understanding is that Sentient Futures is primarily focused on field-building, with a view to supporting interventions as they emerge over time.

One thing I’m still trying to get a better grip on is how this translates into impact on animals, and ideally, on what timescale. I’ve had similar questions when thinking about wild animal welfare more broadly: when does investment in building a field start to produce concrete outcomes that benefit animals? 

In the AI x animals case, it seems slightly more pressing because of the time-sensitivity point. I’m trying to reconcile the idea that 'this is urgent' with an approach that is upstream and preparatory.

I’m also conscious that most of what I’m seeing is the public-facing layer, and you mentioned that a lot of the communication is happening in more private or high-context settings; so it may be that the picture looks more abstract from the outside than it does from within. 

Thanks for the invite. I (edit) was going to join the showcase, but my kid is sick. I hope it goes well. 

Thanks, Martijn. Yes, that's another point of confusion; someone would need to pick up the tab for the added costs associated with better welfare. 

 

Thanks for engaging, Constance, this does answer some of my questions.

I take your point that the window for the EU AI Act was short and that missing it would have meant missing it entirely. Are there any other win-win situations have been found and packaged so far, beyond the EU AI Act?

The BOTEC is a careful piece of reasoning, but it's a model that compares two categories of work without specifying what either involves in practice. It doesn't answer my questions about what concretely gets done, and how animals stand to benefit. 

Your homepage describes Sentient Futures as existing to ‘identify the leverage points for weaving welfare into the core of future systems and cultivate the foundational community needed to activate them.’ A casual reader can’t translate that into what work is happening in real terms. 

You acknowledged the communication gap yourself; I wonder if that's where some effort could go, given you're trying to build a field that hinges on persuading people of the gravity and impact of changes that need to happen now, with time-sensitive urgency.

Oh, no. I had no idea they could be this bad. And I'm speaking from experience... is anybody working on this? 

'When EA becomes cool, people who don't care about EA show up to secure money and status. When EA becomes cool, that reputation needs defending, which is often corrosive to truth.'

I think that's very well said.