@simon Early in my studies I read a quote (in an article about this effect) from an encyclopedia article in the 50s. It described the notion of "machine intelligence" and gave as an example a vending machine that could give exact change.

I've been trying to find that quote for years, but I never managed.

@simon Maybe the issue isn't the fence-post-movers (people saying "this isn't intelligent yet"), but the fence-post-planters: i.e., those boldly offering a definition of intelligence, usually one that depends on performing a given task.

The "is it AI yet?" game may be decades old now, but the game of defining measures of intelligence that turn out to be naive and/or racist is much older, as Stephen J. Gould describes in The Mismeasure of Man.

@simon This reflects the difficulty of defining "intelligence". When a previously-unsolved problem is solved via a method that is clearly ad hoc, the field naturally moves the goal posts. Problem-by-problem, we discover what is easy and what is hard for computers.

@simon I remember when the idea was the computer couldn't learn. That one never made sense to me at all, even on simple computers you can write a program that seems to learn as it goes.

Being introduced to computing by terminal and then PC, I didn't understand that up to that time, most programmers had probably been working on batch processing of data, having put the human "computers" out of work.