I have a few ideas mulling in my head that I’m yet to decide if it would be useful. I’m unsure about posting them as not sure how popular the ‘listicle’ format would be as opposed to my normally very long and detailed posts/comments. These are:
Title: Top 5 Lessons Working in Frontline AI Governance Summary: Things I’ve picked up regarding AI, risk, harms etc working in-industry in an AI governance role. Might end up being 10 things. Or 3. Depending how it goes.
Title: Top 5 Tips for Early Careers AI Governance Researchers Summary: Similar to the above, things I wish I had known when I was an ECR.
Title: Why non-AGI/ASI systems pose the greatest longterm risks Summary: Using a comparison to other technologies, some of my ideas as to why more ‘normal’ AI systems will always carry more risk and capacity for harm, and why EA’s focus on ‘super-AI’ is potentially missing the wood for the trees.
Title: 5 Examples of Low-Hanging Policy Fruit to Reduce AI Risk Summary: Again, a similar listicle-style of good research areas or angles that would be impactful ‘wins’ compared to the resources invested.
I think listicles would be a great style of post for Draft Amnesty (if you’re interested). I’d be interested to see any of the listicles, and your 4th idea would be great to see as a more fleshed-out argument (though it is another one which could be a quickly stitched-together take to post on Draft Amnesty, possibly with a request for feedback in the comments before you do a full draft).
I have a few ideas mulling in my head that I’m yet to decide if it would be useful. I’m unsure about posting them as not sure how popular the ‘listicle’ format would be as opposed to my normally very long and detailed posts/comments. These are:
Title: Top 5 Lessons Working in Frontline AI Governance
Summary: Things I’ve picked up regarding AI, risk, harms etc working in-industry in an AI governance role. Might end up being 10 things. Or 3. Depending how it goes.
Title: Top 5 Tips for Early Careers AI Governance Researchers
Summary: Similar to the above, things I wish I had known when I was an ECR.
Title: Why non-AGI/ASI systems pose the greatest longterm risks
Summary: Using a comparison to other technologies, some of my ideas as to why more ‘normal’ AI systems will always carry more risk and capacity for harm, and why EA’s focus on ‘super-AI’ is potentially missing the wood for the trees.
Title: 5 Examples of Low-Hanging Policy Fruit to Reduce AI Risk
Summary: Again, a similar listicle-style of good research areas or angles that would be impactful ‘wins’ compared to the resources invested.
I think listicles would be a great style of post for Draft Amnesty (if you’re interested). I’d be interested to see any of the listicles, and your 4th idea would be great to see as a more fleshed-out argument (though it is another one which could be a quickly stitched-together take to post on Draft Amnesty, possibly with a request for feedback in the comments before you do a full draft).