The "programming premise" underpinned much of the intellectual work of the 20th century.
The premise holds that understanding something boils down to being able to describe it within a formal language using rigorous propositions, in principle, executable on a computer.
This was most obvious in the artificial intelligence (AI) community, with the success of logic programming over human chess grandmaster Garry Kasparov, defeated by IBM’s Deep Blue in 1996.
But the unspoken ultimate goal was always to understand and simulate ourselves, by accumulating a sufficiently comprehensive and vast corpus of axioms and inference rules, that could in principle be compiled into code and run on a computer.
Solving smaller problems, such as chess, were proofs of the larger concept.
Thus the programming premise was also tacitly implicated in much of the 20th century’s intellectual work outside the AI community.
From logicians inventing logics by the score (predicate, modal, fuzzy) to philosophers who often seemed to see their vocation as writing down what were, in effect, algorithms that could enable machines to make the same fine distinctions in ethics, aesthetics and politics as we humans.
Intellectual workers in the humanities could feel they too were contributing to the growing corpus of computable rigorously expressed and encoded knowledge, perhaps even to its eventual awakening to sentience.
But as it turned out, chess was the zenith of the programming premise, which has been refuted in the 21st century and Deep Blue was a dead-end.
The 20th century dream of programmable sentience worked for chess, but ultimately fails even for the apparently simpler game of go.
It has been displaced by the purely statistical data-driven non-linguistic approach of 21st century AI’s such as ChatGPT.
We have realised you cannot logic-programme your way to linguistic competence, credible artificial intelligence or sentience in the way we had thought.
What then for the future of that certain style of intellectual work we did so much of back in the 20th century, tacitly inspired by the programming premise, across so many disciplines, from mathematics to logic to philosophy to the arts, ultimately in service of the grand project of definitively writing down ourselves and our knowledge in the form of computable propositions?
It is difficult to avoid the impression that it has distinctly lost its lustre, due to the discovery that intelligence and sentience, whether human or artificial, appears to be illegible.
ChatGPT, for example, just as human brains do, encodes its knowledge not as a corpus of legible formal propositions and rules of inference, but as billions of effectively illegible parameters and ultra-high dimensional patterns within a vast statistical model.
The consequent existential crisis facing many 21st century intellectual workers is similar to that faced by scholars in the portentous centuries leading up to the reformation and enlightenment in Europe.
Then, as now, a tacit long-term grand intellectual project inspired by a core premise came to the end of its tenure.
Back then, the premise was that to understand anything involved understanding God, so the vocation of intellectual workers was for each to add their small brick of reasoning on that topic to the growing edifice of theological knowledge in possession of the one universal church.
The demise of this core theological premise and consensus was associated with significant political and institutional turbulence and dissolution.
Monasteries were dissolved and large blocks of political consensus hitherto co-ordinated by deference to the Catholic Church fragmented into many smaller and warring nation states, each answerable only to itself.
What will the demise of our own version of the theological premise mean for us?
Firstly, promisingly, refuting the programming premise means we also refute the 20th century’s demotion of human sentience to that of a glorified programmable calculator.
Not that the consensus materialist view of sentience has been refuted — we are indeed as entirely and purely biochemical as ever — it is just that sentience has turned out to be illegible, cryptically encoded as trillions of synaptic weights and ultra-high-dimensional patterns that can never be written down and traced through, only lived and experienced.
The great Socrates seems to have anticipated this discovery over 2000 years ago for, although he thought, conversed and served his city of Athens much, he pointedly wrote nothing down.
Perhaps, after the initial turbulence of this new enlightenment has subsided, our political and intellectual institutions will have reconstituted themselves along more Socratic and democratic lines, with much more thinking, conversing, doing and political engagement, with much less writing things down.
• Alan McCulloch is a Dunedin former bioinformatics software engineer, familiar with the use of AI techniques in the biological sciences.
