The Unwritten Pairs

AI & Honesty

A model cannot lie in the human sense, because lying requires knowing the truth and choosing to conceal it. It cannot be honest for the same reason. Frankfurt’s third category — indifference to truth rather than opposition to it — fits the architecture uncomfortably well.

Already well covered

This is the most thoroughly worked angle in the whole field. Hicks, Humphries and Slater made the argument at length in 2024 under the title “ChatGPT is bullshit”, and dozens of essays have made it since 2023. Anyone writing it now is entering a full room, not opening one.

AI & Romance

A sculptor falls in love with the figure he carved and the gods animate it. The structural parallel to shaping a conversational partner through your own prompts and preferences, and then being moved by what you shaped, is close to literal. E. T. A. Hoffmann’s automaton Olimpia, in Der Sandmann (1816), is the darker second anchor: everyone can see it except the man in love.

AI & Democracy

Tocqueville’s soft despotism is a tyranny of comfort: citizens do not lose their freedom, they stop exercising it because everything is taken care of. Dostoevsky’s Grand Inquisitor is the sharper anchor and the harder essay — people do not want freedom, they want bread, miracle and authority, and a system that tells you what to think can now supply all three at scale.

Already well covered

The hooks are obvious and the arguments are well trodden. Anything written here has to get past a large existing literature on algorithmic governance and deliberation.

AI & Education

Thamus objects to the invention of writing on the grounds that it will produce the appearance of wisdom without the reality, and weaken the memory it claims to serve. The oldest recorded complaint about a cognitive tool, aimed at the one we now consider constitutive of education. If struggle is where learning happens, what does frictionless access to answers do to formation?

AI & Language

Models trained on an English-dominant corpus exert a gravitational pull towards Anglo-American conceptual frames, and that pull is now inside the drafting of documents in every other language. The essay would need care: strong linguistic relativity is not well supported, and the interesting claim is about institutional vocabulary rather than about thought itself.

AI & Trust

Not the technical question about detecting synthetic media but the political one Arendt was already asking: what happens to a common world when factual truth becomes a matter of opinion about sources. Her subject was lying regimes; the new case is a public that cannot distinguish testimony from fabrication even when nobody is lying.

AI & Craft

When the cost of producing a competent artefact approaches zero, what happens to the cultural standing of making one by hand? Morris was answering the same question about industrial production and got an answer — the Arts and Crafts movement — that was beautiful, influential and economically marginal. The essay would have to reckon with that record.