Crossposted to Less Wrong
‘Like most of life’s dangers, the dangers of artificial intelligence are not easily identified from the armchair.’
J Dmitri Gallow
Epistemic status
I’m not an expert in this field, and herein I make some sweeping generalisations educated guesses about the overall...
TL;DR: EA has traditionally deprioritized climate change as a cause area. But a new report from Rethink Priorities, an EA research institute, suggests that specific Earth systems risks driven by climate change could have mortality impacts comparable to major global health burdens, on top of the known damages of climate change.
The report finds that research and governance work on ‘Earth Systems Stabilization’ to establish whether safe options to address th...
Hi friends! I just a wrote a post on Substack to share some of my thoughts on the prospects of mech interp and CoT usefulness for safety/alignment. It’s inspired by the 80,000 Hours podcast episode with Neel Nanda on the same subject.
https://open.substack.com/pub/mostlyharmlessmachines/p/thinking-machines?r=pm1lz&utm_medium=ios
Would love your thoughts and feedback.