Neither the hopes nor the fears associated with the widespread deployment of recursive self-improvement in generative AI models are likely to be fully justified. In our species, the human species, this recursive self-improvement has given us Mozart, Kant, and Einstein, M.K. Gandhi, Jonas Salk, and Fannie Lou Hamer. It’s also given us Hitler, Stalin, and Mao, Elon Musk, Steve Bannon, and Donald Trump. Viewed from a moral perspective, it’s hard to imagine that any species of artificial intelligence—should that be what these models are actually leading to—could do either any better or any worse than we have.
What’s different about the artificial intelligences our techbro advocates now believe to be imminent are the scale and time frames involved in their evolution as opposed to ours. We were very slow, the machines are very fast. There are more of us and we’ve been around a lot longer, so like it or not, the establishment of a peaceful coexistence between our two species will likely be up to us. As we’ve not been very good at this when dealing with each other, we’re going to have our work cut out for us to come to a detente with this new species that may well prove to be much quicker to understand us than we are to understand them.
NB: Prayer won’t help with this quite unprecedented dilemma. These new creatures, unlike our gods, won’t be figments of our imagination. Like it or not, we’re going to have to think.
