https://forum.effectivealtruism.org/posts/cfn2MMEmpnGjTWpAw/ai-winter-season-at-ea-hotel

AI Safety Tactical Opportunities Fund
@ampdot

(Ops manager David posting on behalf of Greg)

1.) Here are some highlighted outputs from CEEALAR - due to the disruption caused by covid and related staff turnover, we've only recently re-established contact with our alumni and begun to capture more of our successes, thus there are currently large gaps in our known outputs (https://docs.google.com/document/d/1D8zOegcciXgm0B9WBGsV1jAruv_KJvNhwbAO8on3kgU/edit?usp=sharing)

2.) Related to the above, our theory of change is that the work alumni go on to do after their stay with us is where the most impact occurs. That we form a critical part in the journey of aspiring EAs - affording them the opportunity, community and resources necessary to give them the best foundation from which to move the world.

Thus, for example, there have been many people for whom CEEALAR afforded the opportunity to transition into machine learning and AI Safety, and of those many are now working in the area - some of those we know about: one as an engineer at Deepmind, one as an assistant at Anthropic, one as a contractor for OpenAI

I apologise for the lack of concrete data at this stage, but of the ~200 past guests we know something about: at least half are still working in EA cause areas, of those I'd estimate at least a third had no counterfactual alternative to CEEALAR and would not have been able to start/continue working in EA causes areas otherwise, and roughly 10 are in high impact roles (e.g. deepmind engineer).

Turning this into a success rate depends on how you measure success, but relative to the most comparable alternative - funding independent EA researchers work at ~3x the cost - it certainly seems like a meaningful amount of impact for your $.

Apologies again for the lack of specificity, this will be remedied before we apply for funding again in future (assuming we get another ~12 months runway).

3.) I think it's unlikely that anyone missed out by coming to CEEALAR, the opportunity cost of being in Blackpool is dwarfed by the opportunities we provide. And indeed while Blackpool is often criticised for being Not-Oxbridge or Not-London, it is far closer to those places than many of the other places from across the world that our guests are arriving from - staying here affords many of our guests access to these locations that they would not otherwise have had.

3b.) In terms of community health, CEEALAR's community is one of our proudest achievements, and it is something we nurture and are fiercely protective of. Other EA community projects with much more pedigree and less history have had far more serious issues in this regard than we have had in our entire 5 year history.

In fact, for the majority of our serious failures those affected have been staff rather than guests - burnout has historically been a problem for those who have carried the torch of CEEALAR and faced adversity on its behalf.

@Cillerwhale @MarcusAbramovitch

(This is ops manager David posting on Greg's behalf)

Yes we're in Blackpool, a 3-4 hour (depending on the whims of British public transport) train journey from London.

It's definitely a less desirable city to live in, but I think most of our guests find CEEALAR itself to be an incredibly nourishing environment.

We provide accomodation, coworking spaces, communal vegan meals (and I'll go out on a limb here and say that even in London you'd be hard pressed to find vegan food this delicious), coaching and support, and host various talks and retreats which our guests benefit from.

Nowhere can EAs get so much for so little $. In fact, to the best of my knowledge, CEEALAR is unique - both in our mission to onboard and upskill promising EAs from diverse cause areas (we're also technically one of the oldest and most active AI safety organisations - between a third and half of our guests at any one time tend to be working in AI Safety) and in the permanent physical EA community we've nurtured here.