paulfchristiano(Paul Christiano)

Where I agree and dis­agree with Eliezer

What failure looks like

AI al­ign­ment is dis­tinct from its near-term applications

My views on “doom”

Another (outer) al­ign­ment failure story

Thoughts on the im­pact of RLHF research

Self-driv­ing car bets

ARC’s first tech­ni­cal re­port: Elic­it­ing La­tent Knowledge

Thoughts on re­spon­si­ble scal­ing poli­cies and regulation

Hiring en­g­ineers and re­searchers to help al­ign GPT-3

Thoughts on shar­ing in­for­ma­tion about lan­guage model capabilities

An­nounc­ing the Align­ment Re­search Center

Im­pos­si­bil­ity re­sults for un­bounded utilities

IMO challenge bet with Eliezer

Prizes for ma­trix com­ple­tion problems

Se­cure homes for digi­tal people

My re­search methodology

Prizes for ELK proposals

Mo­ral pub­lic goods

AI-Writ­ten Cri­tiques Help Hu­mans No­tice Flaws