The Summer Everything Seemed Possible

In 1956, a program called the Logic Theorist attempted 52 mathematical theorems and proved 38. One proof was shorter than the version in its source book. A machine had carried out work that looked like formal reasoning.

Look at what it was actually doing and the magic thins out. There was no insight involved, no moment of seeing. The program was trying things, one move after another, keeping what led somewhere and abandoning what did not. Done fast enough and thoroughly enough, that looks a great deal like reasoning from the outside.

Which raises a genuinely hard question. If you cannot tell the difference from the outside, is there one?

The people in the room that summer had an answer, and their answer was no, not really, and we are nearly finished. They gave the field its name and predicted the rest would follow within a generation. They were not foolish, and the thing they had built was real. What they had not yet met was the reason trying-everything stops working, which is not a matter of building a faster machine.

That reason has never gone away. It is why a machine could take the world chess title decades later and change surprisingly little about what machines understood.