AI is disruptive and, importantly, discontinuous.
The most instructive moment came in 2016 when DeepMind's AlphaGo faced Lee Sedol, the finest Go player of his generation. Over 200 million people watched. AlphaGo won.
The win was impressive. How the machine won was the stunner, as its “37th move” became the stuff of legend. The machine placed a stone where no serious player would consider (think punting on first down). The commentators went silent, then called it a mistake. Sedol, in shock, stood up and left the room. But it wasn’t a mistake. It was, as one champion later put it, beautiful…and, importantly, a move drawn from outside 5,000 years of accumulated human wisdom.
The machine wasn't imitating us better; it was thinking somewhere we had never been. Discontinuity.
Yet too many of us, in predicting AI’s impact on jobs, commerce, and society, get things terribly wrong. Why? Extrapolation. Drawing a straight line from what came before and assuming tomorrow is an automated version of that.
Look to jobs. This is where the academics keep whiffing, forecasting a jobaggedon that never comes. 13 years ago, Oxford professors Frey and Osborne estimated that 47 percent of U.S. jobs would likely be eliminated, and that we'd now be deep into such job destruction. (Oopsies.) They extrapolated from the tasks machines could already do. Smash cut to today: unemployment is below the natural rate, and in the sectors where AI is hitting the hardest (e.g., software devs) employment is still climbing.
Look to capital markets. We’ve now endured several years of breathless predictions of a market bubble, soon to burst. Yet through wars, election cycles, changes in administrations, etc., markets remain near record highs.
Why? Because the invention is starting to invent. Not just taking from the old, but creating the new.
We experienced this with the assembly line. Did the tractor wipe out 90%+ of farming jobs? Yes. Did it also seed entirely new industries that helped drive U.S. GDP up more than 20-fold (in real terms) since 1929? Also yes.
Why does this matter to you? Because the executives who thrive won't be the ones who simply extrapolate. They'll be the ones who learn to recognize Move 37 when it lands on their board.
The week's most important AI news, curated for you:



















