I'm confused as to why the survey gives concrete examples of issues in AI ethics, but leaves speculative risks abstract "advanced AI could come and pose significant danger to humanity." For the latter it seems simple enough to say "help rogue actors design and deploy weapons of mass destruction, or strip humans of control over the world's resources in AI systems' pursuit of their own goals"
Hi Matt! Thanks for your comment. Replying here on Jamie Elsey's (first author of this research) behalf:
'Thanks Matt - we agree the specific framing or examples could make a difference to responses. In this instance, the precise wording of the statements originally came from an external request, for whom these specific framings were of interest/informative, and we have continued with them so as to see possible changes over time. There is some work on more specific forms of risk, for example, our work with Gruetzemacher and colleagues here (https://arxiv.org/pdf/2406.06199), or extinction risk here (https://rethinkpriorities.org/research-area/us-public-perception-of-cais-statement-and-the-risk-of-extinction/). We’re aiming to run another wave of Pulse soon, so please feel free to reach out to surveys@rethinkpriorities.org if there were related or alternative questions you think would be very useful to know, and we may be able to include them.'
I'm confused as to why the survey gives concrete examples of issues in AI ethics, but leaves speculative risks abstract "advanced AI could come and pose significant danger to humanity." For the latter it seems simple enough to say "help rogue actors design and deploy weapons of mass destruction, or strip humans of control over the world's resources in AI systems' pursuit of their own goals"
Hi Matt! Thanks for your comment. Replying here on Jamie Elsey's (first author of this research) behalf:
'Thanks Matt - we agree the specific framing or examples could make a difference to responses. In this instance, the precise wording of the statements originally came from an external request, for whom these specific framings were of interest/informative, and we have continued with them so as to see possible changes over time. There is some work on more specific forms of risk, for example, our work with Gruetzemacher and colleagues here (https://arxiv.org/pdf/2406.06199), or extinction risk here (https://rethinkpriorities.org/research-area/us-public-perception-of-cais-statement-and-the-risk-of-extinction/). We’re aiming to run another wave of Pulse soon, so please feel free to reach out to surveys@rethinkpriorities.org if there were related or alternative questions you think would be very useful to know, and we may be able to include them.'