AI

When a model does intellectual work for almost nothing, the money that used to pay for it mostly stays in the buyer's pocket. Of what is still spent, the surest profit goes to whoever sells the chips.

Imagine planning an event with an AI assistant. It produces a venue shortlist, a schedule and an invitation. Then you change one premise: children will be coming.

The most capable AI systems now write out their reasoning in plain language before they answer, and a good deal of the effort to keep them safe consists of reading what they write. That looks like a choice.

When translation becomes nearly free, the obvious change is that everyone can read everything. The less obvious one is that being read, and being judged, still happens in a few languages.

Many of the mistakes may start in the record, before any model reads it. A guitar's serial number can belong to several manufacturing histories at once, and the detail that says which one is usually not written next to it.

Anyone with the same tools could produce pieces like these. What the tools will not supply is the pull against their own defaults, and that is where the person behind them shows up.