Question in Brief

The popular usage (I'm not sure of exact technical usage), limits the term "artificial intelligence" to only the "high-end" tasks; as if AI has something limited to "high-end". But "how much high" complication is required to define AI?

Same question Elaborated

We commonly consider "highly complicated" tasks/algorithms/computers to have "AI", say face transformation to age change, Applying painting-effect & assuming 3d rotational views on photos, breaking CAPTCHA codes, optical character recognition (OCR), Smartphone's owner identification by movement pattern detection through accelerometer, tracking social media user's likes and dislikes & thus matchmaking, etc.

Now, what about much less-complicated tasks/algorithms/machines? Say, could we call a simple pocket calculator to have some "artificial intelligence"? or can we tell that a Pascaline, or an old-day mechanical alarm clock, or a mechanical planimeter has some amount of artificial intelligence?

  • 2
    $\begingroup$ Under the hood, an AI algorithm solves a given task with less computational effort. A chess engine which needs 100 days for figuring out the next move, isn't an AI but a normal brute force search algorithm. If the AI can do the same task (determine the next move) in under a minute on the same hardware, then some kind of AI algorithm was introduced, for example alpha beta pruning. The lowest hurdle to call a software “AI” is, that some sort of performance booster is available which makes the problem solving more pleasant than the normal way. $\endgroup$ – Manuel Rodriguez Jul 18 '19 at 18:08
  • $\begingroup$ Thanks for the edit. Please feel free to merge with dupe $\endgroup$ – Always Confused Jul 18 '19 at 18:23