There is an odd feeling where you recognize something to be all-important, and nobody else around you really seems to get it. This can be true of something like art, or some philosophical idea. But what I am talking about here is of something so undeniable it is hard to understand why others do not see it your way. To the first one could reasonably respond, Different people care about different things—leave them alone. But to the latter, such a response is inside the box, so to speak.
This is the feeling I get with AI. A lot of people use the same words to talk about AI. But most people do not live like they mean it. You can tell because we have not shut down AI entirely yet. Or, at the least, because such a proposal, or alternatives, are not commonplace ideas.
Last month I read Michel Justen’s essay AI progress and the feeling of living in two worlds, and I felt a deep resonance with my own life. In one world, I am painfully aware that each day might be my last, and that the world as we know it is about to change forever. In the other, I watch celebrate birthdays, watch movies with friends, and put money into my 401k. Why is there this mismatch? Why do we acknowledge that AI will likely transform the world past recognition, and then invest in our retirement accounts?
One guiding principle for me has been to take the same actions as in the good world, even in the bad world. Even in the bad world, there is a point to it. But why, exactly? This feeling seems to be common even in people who don’t think the above is true.
Perhaps it is partly about a true uncertainty. We are not sure how things will work out, so we do things beneficial to both possible eventualities. And even if AI is radically transformative, it’s not clear how that will work out, or if the mundane tasks of a normal world would truly be worthless.
But that doesn’t seem quite right. I, at least, have a very high certainty that AI will be transformative, and even catastrophic. And yet I still save money, invest in my retirement and my personal investment account, work on developing skills that will not come into fruition until after the AI event horizon.
It is probably more like this essay’s title. It is like living in two worlds. Of course, we all fundamentally live in the same world, but that can feel very different for different people, or for the same person in different moods and contexts.
I suspect that the claims of radically transformative AI are just too hard to believe, in a number of senses. First, they are simply outside the usual realm of possibility. We are surrounded every day with a thousand reasons to expect otherwise. Second, it’s hard to see concretely how this will work out. The world mostly feels normal now. The problem is that AI works on an exponential curve of change, and we are not very good at feeling exponentials. By the time life feels weird, it will already be too late.
Because of this, while I might nominally believe AI will change everything, it can be hard to feel that all the time. And it is not clear to me that feeling that all the time is even beneficial for a happy life. Perhaps it is. Ben Kuhn has written on staring into the abyss as a core life skill. We ought to welcome bad news. Still, it is hard. There is also the desire to Not Be an Alarmist. We are worn out from alarm. Still, I have the nagging feeling that living in light of this, feeling it constantly, is the key to living well. It is downstream of memento mori. It is easy to forget. And when forget we do, we are merely drifting by in the sands of time. Only when we remember, blighted though we are by the desert sun of truth, can we pull ourselves up and start our pilgrimage to the promised land.
This feeling is probably true for the world as it is for the individual. Only the TAI-realist voice is much smaller for the world. I suspect we will only get anything done if we eliminate this confusion and learn to steer away from the proverbial black hole, blinded by ignorance as we are. We need people to wake us from our slumber.