Birrell,

I am currently working on a new piece about why we do not need to fear the current incarnation of AI.

The basic idea was taken from an off-handed comment made by Clive Thompson. He was talking about a recent horror movie about AI (called M3gan) and pointed out that the least believable part of it was that the titular AI slaughter-doll was capable of moving like a human.

In reality, engineers find it incredibly tough to create a machine capable of consistently executing motions that humans find easy, under all of the varying possible conditions where those motions might occur.

Humans, and animals more generally, seem to have a series of powers (like the ability to generalize and deal with edge cases and rare incidents) that put AI to absolute shame. But these things feel so routine to us that everyone has always taken them for granted as "simple" things.

The story, then, is that in trying to create machines that mimic what we have classically considered the greatest of human capacities -- logic, creativity, reason, insight, connection of disparate domains of information, etc... -- we have actually left out a lot of systems that are far more profoundly difficult, assuming that they were simpler and we could solve those easily.

But I think we will find out that they are very, very difficult indeed. And in perhaps the greatest blow to our ego we will find out that these much-more-complex systems weren't the things that made us human. They were the things we shared in common with all animals.

Or, more simply put: We have now succeeded at the very small task of creating a machine that can make art, and must tackle the much larger and more complex task of creating a machine that can want to make art.

Until we get that I think we have nothing to fear from AI.

Best

J

P.S. - My current article may or may not materialize, but I wanted to tell you about it because it was inspired in part by your comment and by some of the others I received after writing this piece.