Every major man-made problem keeps coming back because the values of the people meant to fix those problems remain unchanged, and the systems producing those values never changed either.
Some AI leaders are building AGI to solve the fundamental problems in the world.
I’m curious, how could AGI solve every major man-made problem if the root cause is people’s values and world systems? Even if we solve the technical problem of making the AI do what we ask, it won't matter if we are still asking it to serve broken human values and systems.
If technical alignment is a very hard problem to solve, why can't we just use the AI systems we have now to help us improve people's values and world systems?
We could even find a way to inform the technical alignment problem as we focus on improving people’s values and world systems. By practicing with the AI we have today, we might learn how to better describe and build human values into the smarter AI of the future.
I’m curious, how can we get frontier AI organizations to prioritize using current AI to help improve people’s values and world systems as a way to support the overall goal of a safe and useful AGI that benefits humanity?
Every major man-made problem keeps coming back because the values of the people meant to fix those problems remain unchanged, and the systems producing those values never changed either.
Some AI leaders are building AGI to solve the fundamental problems in the world.
I’m curious, how could AGI solve every major man-made problem if the root cause is people’s values and world systems? Even if we solve the technical problem of making the AI do what we ask, it won't matter if we are still asking it to serve broken human values and systems.
If technical alignment is a very hard problem to solve, why can't we just use the AI systems we have now to help us improve people's values and world systems?
We could even find a way to inform the technical alignment problem as we focus on improving people’s values and world systems. By practicing with the AI we have today, we might learn how to better describe and build human values into the smarter AI of the future.
I’m curious, how can we get frontier AI organizations to prioritize using current AI to help improve people’s values and world systems as a way to support the overall goal of a safe and useful AGI that benefits humanity?