Can the flame of mathematics, the historic logos which was recognised as the Queen of the Sciences —and as the most authoritative category of human knowledge— for more than two millennia, survive in today’s fevered, over-hyped promotion of AI, based on neural networks? Mathematics has been, in effect, by-passed by the computer establishment, which routinely speaks as if its successful projects “have nothing to do with maths”  and are mainly down to the use of clever modern electronics. The unique insights which earlier mathematicians were able to bring to bear onto abstract problem-solving have been exploited massively in common software. But simultaneously they have been virtually airbrushed-out of the accounts put out by Silicon Valley. This malpractice seriously harms the credibility of maths in schools.

The maths education system is being stripped of its motivating thunder. 

They got away with this malpractice … mainly because the maths establishment chose to cut a pathetic figure with their emphasis on unworldly, ornamental, in-house puzzles.

Now the computerists are repeating the same trick with so-called AI. (They call it ‘Intelligent’ though it is actually copycatting. It doesn’t do any genuine thinking, at all. It merely hashes together stuff which sounds —most of the time— like sense.)

This has spawned a heavily financed bandwagon which seems to have emerged    —and become fiercely defended– because leading actors in Silicon Valley were gobsmacked in 2022 when their machines started outputing paragraphs of grammatically correct language… This sounded exactly like human judgments! So the gurus of Silicon Valley jumped to the conclusion that they were more powerful than human judgments! If these experts had been more savvy about how language works, they would not have fallen into this trap: they would have realised that it was a “hit or miss” mishmass which looked good most of the time, but which wasn’t reliable.

On July 16^th The Telegraph newspaper printed an article which contained serious, devastating evidence that AI is a bubble destined to burst… because its supporters have committed themselves to a flawed project, foolishly, on a massive scale.  It is now said that more than a trillion US dollars has been invested in this half-baked exercise, which is based on the naïve assumption that the current deficiencies of AI-based “judgments” will soon be “sorted out”.  A sign that enthusiasm for AI has gone over the top is that some AI firms have decided to try to apply it to weather forecasting.  If so, they are evidently not aware that the Earth’s atmosphere is a diverse system of astronomic proportions, and that all the tricks of AI are not going to make the slightest difference to the hopelessly difficult task of modelling it accurately. Meta is reported to have tried to lure especially talented AI experts with sums as large as $200 million. This is a clear sign that a degree of panic is beginning to surface.

AI has, so far, managed to get-by without paying for the content it uses to “train” its large language models, and lots of ordinary commercial businesses who use it have not yet woken up to the fact that human experts will be needed to check all AI outputs —to ensure that they are not hallucinating.  Once these costs are factored-in, the prospect of huge benefits arising from AI are likely to wither.

A PRIZE OFFER, THE RESULT

To mark the fifth-year anniversary of the author’s so far unchallenged “bare hands” analysis of Fermat’s “Last Theorem”, the author decided to offer a reward of £3,000 to the first person who could find a self-evidently exposition-derailing mistake in the on-line argument.  He or she would have to recruit two independent colleagues to check and confirm his or her claim —they will each receive £1000.

It is now August 1^st and  no such counter-exposition has been received.

The exposition in question is on blogs 1 and 2 of this website.

[However a new version of the offer will now take-over. The bounty is to be increased to £5,000. It is now being offered to the first person who can find a seriously disabling fault in the putative proof.  He or she will also have to recruit two checkers, who will verify the counter-thinking, and will receive £1000 each if such a disproof is found.  The new deadline is Christmas Day 2025.]

This is not rocket science… or even really difficult cutting-edge maths:  it is really elementary… and in terms of skills it barely needs more than double maths A-level in order to follow the reasoning.  Its validity, or otherwise, urgently needs to be established.

The author now intends provisionally to adopt the default assumption that his bare hands reasoning on the Fermat Problem <<Is most likely to be correct>>.  This is not a project to undermine Wiles’ brilliant advanced modern proof of Fermat: on the contrary.  Its object is to remove the intense embarrassment associated with more than 350 years of the mathematic community’s seemingly abject inability to reconstruct what Fermat must have had in his hands in the 17^th century. To enter for the extended prize: send your name, putative disproving reasoning, and confirming colleague names,  to:  per4group@gmail.com .

PS:  The P E R Group’s new BULLETIN can be found on-line at:  PERtinence.co.uk/pertinence-news-1/

CHRISTOPHER ORMELL around August 1^st 2025. chrisormell@aol.com