We keep arguing about “AI” as if it has a stable definition.

[mountainhermit@mastodon.social]

We keep arguing about “AI” as if it has a stable definition.

It doesn’t—and it won’t—because “intelligence” is a moving target, and every definition depends on where someone’s work (or fear, or incentives) intersects with the tech.

That’s why debates keep collapsing into talking past each other: we’re not using the same word.

#AI #TechPolicy #MediaLiteracy #DigitalCulture #PlatformPower

[teledyn@mstdn.ca]

@MountainHermit

This our fault, those of us who help define the field, and I apologize. We were duped by science fiction into thinking the Turing Test was valid, it was pompous and overstated but reflected our naive ambitions and the term gained funding attention so who could blame us?

The CORRECT term would be Applied Statistics

#WhatsInAName

[77nn@goto.77nn.it]

@teledyn @MountainHermit

AI is just a catch-all word. It's not the meaning of *intelligence* that is a moving target, but the definition of what is inside AI that is. It could have been named "applied statistics" and include the current dozens of technologies that are in it.

Translators are AI, Map Navigation is AI, Search algorithms are AI. A lot of these technologies are pretty awesome and belong to our normality.

The limit between "automatic" and "ai" is always moving. As things get more complex any previous technology moves from being "AI" to being simply some kind of "automation". This moves the target.

Today, instead, we are faced with an inconsistent technology (namely LLM) that claims to be the ouverture for AGI, the Graal of computing. The claim is false, and the technology inconsistent as it doesn't have any specific use case.

They promise it is being the "all use cases that you may think of", but if you coded a little bit you know that this is bullshit.

Bingo.

Guess why the first jobs to be replaced are the creative ones.