This was a very illuminating newsletter. It is nice to hear a diversity of perspectives on alignment.
Second, we don’t know how difficult AI safety will turn out to be; he gives a probability of ~10% that the problem is as hard as (a caricature of) MIRI suggests, where any design not based on mathematical principles will be unsafe. This is especially true because as we get closer to AGI we’ll have many more powerful AI techniques that we can leverage for safety.
How accurate is it to say that MIRI believes a mathematical solution to the alignment problem is the only solution? Does MIRI think that without a formal proof of an AGI’s safety, it will cause human extinction?
This was a very illuminating newsletter. It is nice to hear a diversity of perspectives on alignment.
How accurate is it to say that MIRI believes a mathematical solution to the alignment problem is the only solution? Does MIRI think that without a formal proof of an AGI’s safety, it will cause human extinction?