AI Safety Support conducted surveys to people they had worked with previously.
Actually, most people where not people we engaged with before. We spread the survey as wide as we could.
Half of them came from Rob Miles audience, or that’s my estimate based on timing of responses.
But thanks for analysing our data!
I never got around to it, since I had to leave AISS a few months later, under pretty bad circumstances (which was not the fault of anyone working for AISS!)
Another interesting result of that survey was that a substantial fraction of applicants wanted some type of help or resource that did already existed but which they were unaware of. I can’t remember now what percentage of applicants this was, but I think it was something in the range 20%-70%
There’s a minor mistake in the post
Actually, most people where not people we engaged with before. We spread the survey as wide as we could.
Half of them came from Rob Miles audience, or that’s my estimate based on timing of responses.
But thanks for analysing our data!
I never got around to it, since I had to leave AISS a few months later, under pretty bad circumstances (which was not the fault of anyone working for AISS!)
Another interesting result of that survey was that a substantial fraction of applicants wanted some type of help or resource that did already existed but which they were unaware of. I can’t remember now what percentage of applicants this was, but I think it was something in the range 20%-70%
I wrote about it here:
AI Safety Career Bottlenecks Survey Responses Responses
Hi Linda—Thanks for the clarification on the source of your data. Updated with the corrected information.