What we don't know, and why saying so is a good sign

On many central questions, the honest answer is "we don't know." That's neither a dodge nor an admission of weakness: it's the actual state of the subject, and knowing which questions remain open protects you better than any certainty.

The genuinely open questions

• Will progress continue at the same pace? Nobody knows. Everyone's bets are bets, not projections.
• What happens inside the model? We know how to build it and measure it, but we can't explain a particular response. An entire field of research is devoted to this, with partial results.
• How do you guarantee behaviour? We can make a behaviour highly likely. We can't guarantee it.
• What effect on employment in ten years? Serious studies contradict each other, because they measure different things.

Why nobody says so

Because uncertainty doesn't sell. A headline that makes a claim gets picked up; one that hedges doesn't. An expert who asserts gets invited back on TV; an expert who doubts doesn't get called again. The mechanism isn't a conspiracy, it's a selection effect: bold claims circulate better, in both directions.

Consequence: what you read most often isn't what's most true, it's what's most shareable.

The sign that distinguishes a good source

A good source states what it doesn't know, distinguishes what it has verified from what it has been told, corrects itself when it's wrong, and gives you enough material to contradict it.

A source to avoid asserts everything with the same confidence, never gets it wrong publicly, and predicts dates.

One final warning, and it applies to this site too

This lesson was written on a given date. Some open questions will be settled, others will emerge. A course that claimed otherwise would deserve exactly the distrust it teaches you to have. Including this one.

Do this right now: pick a source you follow and look for its latest published correction. If you can't find one, it's not because it never gets things wrong.