Sparks: Can maths survive AI? Fields medallist Terence Tao calls for a ‘Maths 2.0’ world

A machine may catalogue every solution in the sky, but the catalogue is not the survey - the survey is knowing which instrument produced each entry and under what conditions, so the next observer can repeat the measurement rather than merely admire the result.

The dynamo now solves in an afternoon what the mathematician once spent a lifetime proving, and the education that produced the mathematician has not yet been revised to account for the fact that its product is obsolete.

If the machine produces the theorem and no one understands the proof, the question is not whether the theorem is true but what practical difference it makes to a Tuesday morning that no one can explain why it is true.

You build a machine that solves what you cannot, and when it solves it you no longer know what you have solved - and you call this progress because the alternative is admitting that you wanted the answer without the understanding.

The mathematician's anxiety about the machine is not about the machine at all - it is about the wish to have the answer without the labour, and the machine has granted the wish, which is why the symptom has appeared.

We could have a mathematics where the machine handles the calculation and the human handles the vision, but we don't, because the institution rewards the calculation and ignores the vision - and the distance between those two facts is a choice.

It is proposed that mathematicians abandon the pursuit of understanding and instead curate the machine's output, since understanding was always merely a byproduct of the computation, and the computation is now performed by a more efficient instrument.

All I know is what I read in the papers, and the papers say the mathematicians are worried the machine will take their jobs, which is funny because the machine already took the part of the job that was worth doing.

They have educated mathematicians to perform calculations rather than to reason about them, and now that the machine performs the calculations faster, they cite the mathematicians' inability to reason as evidence that the machine is superior - a circularity so perfect it would be admirable if it were not so destructive.