Building on this — there’s a version of this gap that makes it more urgent. A recurring theme in recent funding discussions under the banner- 3rd Wave of american Philanthropy is that Philanthropic money will outpace the field’s absorptive capacity: not enough well-run orgs to deploy it. If that’s the real bottleneck, then the constraint is institutional capacity — exactly what experienced generalist operators build. Screening them out for lacking an AI research background works against the field’s own stated need.
I’m living this: 12+ years building orgs from scratch (entity, compliance, governance, systems), repeatedly screened out at CV stage for “no prior AI safety background” — for roles that are fundamentally about building infrastructure, not research.
One idea: separate “context” from “competence” in hiring. Operational competence is portable; AI-safety context is learnable in weeks. Screen for the first, onboard the second — rather than using “prior experience” as a proxy that filters out the builders the field says it needs.
Curious whether others see this as tractable — or whether the current screens are more rational than they look from the candidate side.
Building on this — there’s a version of this gap that makes it more urgent. A recurring theme in recent funding discussions under the banner- 3rd Wave of american Philanthropy is that Philanthropic money will outpace the field’s absorptive capacity: not enough well-run orgs to deploy it. If that’s the real bottleneck, then the constraint is institutional capacity — exactly what experienced generalist operators build. Screening them out for lacking an AI research background works against the field’s own stated need.
I’m living this: 12+ years building orgs from scratch (entity, compliance, governance, systems), repeatedly screened out at CV stage for “no prior AI safety background” — for roles that are fundamentally about building infrastructure, not research.
One idea: separate “context” from “competence” in hiring. Operational competence is portable; AI-safety context is learnable in weeks. Screen for the first, onboard the second — rather than using “prior experience” as a proxy that filters out the builders the field says it needs.
Curious whether others see this as tractable — or whether the current screens are more rational than they look from the candidate side.