1/ Over the past two days Robert Long and I spoke at Anthropic, Google DeepMind, and OpenAI about the case for taking AI welfare seriously. Great conversations at all three - thanks to everyone who organized and attended!
Comments
Log in with your Bluesky account to leave a comment
2/ At each talk, we recommended that AI companies start taking AI welfare seriously through three basic steps: acknowledge, assess, and prepare. While the actions of companies are far from sufficient, they can still help pave the way for a more comprehensive societal approach.
3/ We made these recommendations because we take there to be a realistic chance—far from certain, but likely enough to warrant caution—that AI systems will soon have morally significant interests. In addition, people will increasingly believe they do. Society needs to prepare.
4/ You can find more information about our arguments and recommendations for AI companies in our 2024 multi-author report "Taking AI Welfare Seriously."
Comments
Thanks again to our hosts for a great two days!
https://anatomyof.ai/
https://thebullshitmachines.com/