A quick note, is that looking at the top 5 reading list and the headline conclusions from these, I'm tempted to basically do the opposite and anti-recommend them (at least 1-3 and 5). The zeigeist on awareness and inclination around x-risk messaging in the US specifically has changed in a very major way this year, and all of this message testing is really outdated now. Private message testing is currently giving very different results.

Applications to SPAR Fall 2026 are closing tomorrow Aug 18 EOD Anywhere on Earth. SPAR is the ecosystem's biggest AI safety research program, and it's part-time remote.

We still have many strong projects across AI safety, AI policy, and biosecurity with very few applications^[1], so please consider applying! This round, we also have 18 non-research/generalist projects that people can apply to, and we have much more mentee capacity than previous rounds; we expect to accept around 500 people into the program.

If you have friends who have thought about going into AI safety, spread the word!

1. ^{^}

   To be specific, as of 2:25 PM PT, we had around 67 projects with fewer than 20 applications total!

Both of your points match some of my analysis for early-career people in Chile going into AI safety as well. I think the impact calculus is really similar, and the debate feels similarly unfocused. People have a strong preference for local interventions, but these are generally (though not always) not justified by a strong impact case. It's sad, but I do think most of the strongest opportunities for impact are, for better or worse, in the US, China, UK or the EU.

Thank you for writing this! This is extremely cool.

Kairos (my org) doesn't run a ton of ads, so we don't do this, but one thing we recently started doing is to consistently push all of our funnel analytics data to PostHog, and then connect Claude to PostHog via MCP to pull out a more qualitative analysis of our application funnels for programs. I've thought this has been moderately helpful so far at identifying opportunities and letting us understand where applications are coming from. 

Separately, I recently read this blogpost which goes over a similar workflow that is used by growth marketing at Anthropic. I think this is a pretty good reference for orgs thinking about doing this.

I'm really excited for this! 

mucho éxito!

Hey, I downvoted this and wanted to explain why.

I think there's a real concern under this, but the framing overclaims what the evidence supports. The specific examples mostly have mundane, well-documented explanations: contradictory post-hoc rationalizations are a known consequence of LLMs lacking privileged access to their own sampling (calling it "gaslighting" anthropomorphizes in a way that obscures the actual mechanism), and Japanese rendering failures are downstream of CJK tokenization plus much smaller pretraining-data share, a failure mode shared across Arabic, Hindi, Korean, etc. Reading these as evidence of architectural deprioritization specifically targeting non-Western users requires a comparison condition the post doesn't have. I think it's not an implausible hypothesis, but there's a nuance to be had re. intentionality vs this being the default outcome of the relevant pretraining datasets.

I also don't buy the bridge to existential risk. The argument is roughly: current systems optimize for measured things over unmeasured things, therefore documenting cultural-fluency gaps is alignment work at AGI scale. The first step is just Goodhart and applies to almost any deployed ML system; the second step requires assuming that "renders Japanese poorly" is mechanistically predictive of the failure modes that produce catastrophic outcomes from much more capable systems. I don't think that holds. The threat models that drive x-risk concern (deceptive alignment, power-seeking, loss of human oversight) aren't strongly related to multilingual capability being uneven (especially in image models, which aren't generally capable), and collapsing them probably weakens both conversations.

I'd also gently flag that the methodological pushback (the Apart Research response) gets folded into the theory as further evidence of the pattern. I think this is a bad argument, because it proves too much; it is a fully general defense for these types of arguments, pretty much regardless of what's true. 

There's probably a good paper here about multilingual evaluation gaps, and I'd recommend checking some of the existing work on this topic. One possible direction is trying to make this into a controlled cross-language study with blinded LLM or human raters.

Wishing you well on the project!

I don't believe these things are mutually exclusive. The strongest founders/operators I know want to move the needle in a specific market, and if you want those folks, then it helps to frame the conversation around the problems they're already attacking.

Yeah, this makes sense. I agree that this is another intervention we should consider doing in parallel. 

Btw, happy to see more YC founders around :)