Ethan Mollick, a Wharton professor who studies how people work with new tools, made this the fourth rule of his book Co-Intelligence. It sounds like a sales pitch. It is an expiry date on your opinion.

Most verdicts on AI were formed in one sitting. You asked it for a client proposal in 2023, it invented a figure, and you filed the whole thing under "not yet." That verdict felt like judgement. It was a snapshot.

Mollick's line says the snapshot was taken at the bottom.

The version that failed you is the worst one you will ever meet, and the tool has not stopped changing since. So the "not yet" in your head carries a date you never wrote down, and it passed without telling you.

The cost is not the tasks you still do by hand. It is the judgement you are not building. The person who worked with the clumsy version learned where it lies, how to check it and what to hand it, and they carry that into every better version. The person who waited for the good one gets the tool and none of the instinct.

That gap compounds the way a skill does. Nobody closes it in a weekend.

So today, twenty minutes: take the one task you decided AI could not do, the contract summary, the pricing table, the first draft of the difficult email, and run it again on whatever you can open this morning. Check it as hard as you did the first time. If it still fails, you have a fresh verdict. If it works, you have been paying for a job that ended months ago.