Something exhilarating and terrifying is about to happen.
So far, across its 300,000 years or so, humanity has solved problems, individually, artesanaly, one at a time.
There are periods of faster innovation and periods of slower innovation and sure, there is a general and exponential speeding up of innovative problem solving.
Team research exists, but seldom solves two big problems on the same day.
All of that is within the same paradigm of the human intelligence approach to problem solving.
We’re no longer alone as the only intelligence capable of industrial scale innovation.
Within about a year (mid-2024 to mid-2025), AI moved from silver-medal level performance to gold-medal level at the International Mathematical Olympiad (IMO), with multiple systems (DeepMind's Gemini, OpenAI's models, Harmonic) achieving scores equivalent to top human contestants by the 2025.
AI is solving or very close to solving many of the big problems, from olympiad maths challenges [1.] to rocket motor geometry [2.] or protein folding [3.].
It used to be the case that these problems were monopolised by a tiny number of our most specialised minds, uniquely suited to, interested in and motivated by solving them. There was a queue. It was fed by something as capricious as inspiration and bottlenecked by talent and IQ.
That bottleneck has moved, enabled by the combination of Nonhuman and Human intelligence.
Compute, chip size, cooling and power generation are just as relevant.
Immagine moving from the scarcity of sporadic innovation to the simultaneous invention across disciplines. The big problems in Maths, Physics, Biotech, Medicine and more.
2026 is one of the last years where we will need to imagine that, and I find it deeply shocking at a level which makes Apple's “this changes everything” iPhone slogan absurd.
By Phil Blything
