What is "artificial intelligence"?

“Artificial intelligence” is just a fancy term for “probability modelling”

I’m skeptic towards the term “artificial intelligence”. It has been here for more than 50 years, and it’s not used consistently.

It couldn’t be used consistently since its meaning is so vague. You don’t need to be particularly intelligent to walk on two legs, and yet creating a robot that can walk on two legs (or even on four legs) is in the domain of “artificial intelligence”.

I think that in its beginnings, “AI” meant more or less what the words say—it was the problem of creating a computer that would think as a human. While the task turned out to be unrealistic, it helped create a number of useful technologies, and many of these have kept the misleading name.

Fast forward to today, and you have a great marketing term. “Our app uses artificial intelligence to predict floods” is so much more impressive compared to the more accurate and meaningful “our app uses a sophisticated probability model to predict floods”.

In fact, “probability model” is very often the correct term to put in place of “AI”.

There’s more on “AI” and related terms in a classic article by W. M. Briggs.