Hamming and the future of AI

I’m read­ing The Art of Doing Sci­ence and En­gi­neer­ing by Richard Ham­ming. Yes, the same Ham­ming be­hind Ham­ming codes and Ham­ming dis­tance.

Right on the first chap­ter he cites a well-​known adage about pre­dict­ing the fu­ture:

There is a say­ing, “Short-​term pre­dic­tions are al­ways op­ti­mistic and long-​term pre­dic­tions are al­ways pes­simistic.” The rea­son, so it is claimed, the sec­ond part is true is that for most peo­ple the geo­met­ric growth due to the com­pound­ing of knowl­edge is hard to grasp.
The Art of Doing Sci­ence and En­gi­neer­ing — Richard Ham­ming

Then he of­fers ar­ti­fi­cial in­tel­li­gence as a coun­terex­am­ple, not­ing that AI lead­ers had made pre­dic­tions that:

[…] have al­most never come true, and are not likely to do so within your life­time […]

The book was pub­lished in 1997, based on a course Ham­ming taught at the Naval Post­grad­u­ate School. Most of his stu­dents are likely still alive today. His skep­ti­cism to­wards AI prob­a­bly re­flects the AI win­ter cli­mate of the time. He was right that AI ex­perts had been mak­ing far-​fetched pre­dic­tions for decades. But with hind­sight, those were ac­tu­ally the op­ti­mistic pre­dic­tions. The thing is, AI’s “short-​term” took decades. Now here we are with LLMs, speech recog­ni­tion, com­puter vi­sion, and so on. If any­thing, this con­firms the adage’s truth.

This isn’t a cri­tique of Ham­ming. After all, he knew pre­dict­ing the fu­ture is hard. There is a whole sec­tion about AI later in the book that seems to pro­pose in­ter­est­ing con­cep­tual ques­tions, though I haven’t read it yet. So far, I’m en­joy­ing the book’s goal to think about think­ing and learn­ing for the fu­ture.