The story is computers

Rather than stand apart as some kind of rev­o­lu­tion or rupture, lan­guage models should mostly cause us to reflect on the power of all com­puters, the magic of them, which is this: Here is an engine that can take sym­bolic instruc­tions and make com­plex things happen.

There have been lots of tools in human history, and only a very few of them, starting with the auto­matic loom, have this capability. (There are a few other candidates, fur­ther back … one is Leibniz’s Stepped Reckoner, what a name.)

It’s instruc­tive to imagine a world with lan­guage models but without com­puters; maybe in that world they run on some weird bio-technology — maybe they really are plants, grown on elab­o­rate trellises. In that world, they are still astonishing, but much less useful … because there’s not already this vast auto­matic envi­ron­ment in which lan­guage (the kind called code) becomes action.

This isn’t a paean to com­puters — I think a sig­nif­i­cant part of their auto­matic realm is basi­cally use­less and stupid — but I do want to insist on the con­ti­nuity of the story. It runs straight through, from punch cards to main­frames to per­sonal com­puters to whis­pering agents.

(I realize this is basi­cally a restate­ment of my last post—as you can tell, I’m still thinking about it!)