I have recently visited a follow-up( IMHO/likely) talk of Jan Kulveit. His main hope, IIUC, is that we will be able to put together “humanity agent”, which unlike the name suggests would include current and future GPAIs. The logic of the hope is clever: If the humanity is a shared identity, the powerful AI will not want to change what it( the AI≈humanity) is, thus will strive to persist and ideally empower us, bio-made humans.
The talk was held in Prague and likely should appear on ACSResearch YouTube channel soon.
I have recently visited a follow-up( IMHO/likely) talk of Jan Kulveit. His main hope, IIUC, is that we will be able to put together “humanity agent”, which unlike the name suggests would include current and future GPAIs. The logic of the hope is clever: If the humanity is a shared identity, the powerful AI will not want to change what it( the AI≈humanity) is, thus will strive to persist and ideally empower us, bio-made humans.
The talk was held in Prague and likely should appear on ACSResearch YouTube channel soon.