It reads as reassurance. It is a deadline.

Mollick is not promising you a better tool next year. He is telling you that every judgement you made about what these things cannot do carries an expiry date — and you are still acting on the expired ones.

You made those judgements once, in a single afternoon. It got the numbers wrong, invented a source, wrote something you would have been embarrassed to send. You closed the file and moved the task back to the human column.

That column has not been reviewed since. Six months have passed, which is three model generations.

Nobody sends a notice when the thing you ruled out stopped being impossible.

The people pulling ahead are not the ones with better access. They are the ones who re-test. They keep a short list of what failed and run it again every few weeks — not out of enthusiasm, out of bookkeeping.

Half that list still fails. The half that stops failing is where the week gets shorter.

Your advantage was never the model, because everyone is handed the same one. The advantage is how recently you checked.

So today: take the one task you decided was beyond it — the messy reconciliation, the first-draft contract, the deck nobody wants to build. Twenty minutes, an honest prompt, and a real look at what comes back.

You may find you have been doing by hand a job that stopped needing hands in March.