First you hear Noam Chomsky, as reported by the remarkable Ernie Schenk, calling it “high-tech plagiarism. Then you dial in the “godfather” of the technology, Geoffrey Hinton, who warns us of its dangers. All while the entirety of formerly-FAANG-now-MAANG giants are scrambling as fast as their little coding fingers can fly to build the competition-terminators of their dreams. 

Wherever you land on the boon-or-bullchips spectrum of AI opinions, it’s clear the war of words has been joined.

And who among us can resist a good food fight with all 26 letters of the alphabet soup?  Even if we write it ourselves, sans Chat GPT.

My vote for the best take to date comes from tech CEO Alan Baratz in response to Politico’s weekly question, “what’s a technology you think is overhyped?”

“Generative AI is way overhyped. AI and machine learning are just about pattern matching, and so long as the future looks like the past, you're going to get some benefit out of it. But really, what generative AI is about is pulling out relevant information from a large body of data by doing pattern matching, so in my view it's nothing more than another productivity tool. All the hype around how totally transformative it’s going to be is just that, a lot of hype (emphasis added).

Now, in fairness, Baratz is not exactly playing objective Solomon here. He goes on to say the Feds would be oh so much wiser to invest our tax billions in quantum computing, and, shocking as it may be, that’s exactly what his startup is all about.  

But his point is that a system built on finding patterns in existing datasets, no matter how large, is bound to replicate the past instead of inventing the future. And that’s gospel, at least if the experience in working with some of the earlier versions holds fast.

He’s right: we’re talking process (speed) and productivity (efficiency) gains. Not creativity in its truest sense.

Now, this isn’t to say that there aren’t novel patterns to be discovered in the constellation. You bet there are, and that augurs well for AI as an iterative tool across an amazing range of analytic, predictive, and production possibilities.

But when it comes to breaking molds and intuitive leaps we’ll have to look elsewhere. Maybe upwards, if the other person happens to be taller than you.

S’truth. Hand to HAL.