Strong vs. Weak Artificial Intelligence (AI)

Is a machine that meets Minsky’s Maxim intelligent?

Not necessarily.

Certain machines are built for what engineers call the “4Ds”: machines built for the purposes of performing tasks deemed too Dangerous, Dirty, Dull, or Difficult, for humans to do.

Such machines successfully complete tasks that would require intelligence if done by humans, but that does not mean that the completion of said task(s) makes the machine intelligent or reveals something noteworthy about cognition.

Of more interest may be the distinction between so-called “strong” and “weak” artificial intelligence (AI), which was first introduced by John Searle in his 1980 paper, “Minds, Brains, and Programs.

As Joel Walmsley writes in Mind and Machine, “strong AI” is a hypothesis, or claim that is either true or false.

Researchers focused on GOFAI were attempting to solve the so-called “strong AI problem”: is producing, or re-creating, actual cognitive functions in machines possible?

Unlike “strong AI,” so-called “weak AI” is not a hypothesis, but rather a method of theorizing.

Through building “weak AI,” researchers can not only automate certain tasks (whether those tasks fit under the 4Ds or not), but also learn about cognitive processes.

Generally speaking, researchers building (or theorizing about) “weak AI” focus on the mimicry or simulation of cognitive functions within a machine.

In conclusion, “weak AI” is the method of study, whereas “strong AI” points to the claim that machines possess the very thing being studied: the mind.

Note: Today, “strong AI” is more commonly referred to as Artificial General Intelligence (AGI), while “weak AI” has been dubbed “narrow” intelligence.” While the terms are close enough to be analogous, there are some slight variations between “strong AI” and AGI and “weak AI” and “narrow” AI. Let us know in the comments if you would like to hear more about this distinction, and we may follow up with a post.

Next time you are reading an article about AI, ask yourself: how is the author defining AI, and do their claims concern the “weak” or “strong” conception?

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.