Lines of Thought

01Intelligence & belonging

Who gets to be called intelligent?

“Maybe we are all AI. And maybe we are not, all at once...”

Akomolafe asks what is smuggled into the contrast between artificial intelligence and supposedly natural human intelligence. He acknowledges the disruption of deepfakes and threatened livelihoods, while questioning the claim that wisdom and creativity belong exclusively to humans. His playful possibilities leave intelligence relational and unsettled, rather than declaring that machines are conscious or dismissing their harms.

02Culture & imagination

An AI raised in relationship

In Quartet, Lewis imagines a child growing up with three AIs shaped by different understandings of the world: Kānaka relations to land and family, Blackfoot attention to flux, and an octopus-inspired distributed mind. Intelligence emerges through their ongoing dialogue and shared development. The fiction asks how AI might support situated flourishing without making every mind understand the world in the same way.

03Labour & personhood

The people hidden inside the cloud

Drawing on African feminist accounts of unpaid care, Nyabola asks whose labour disappears when AI is described as data and computation. Domestic care subsidizes paid production; data preparation, moderation and creative work similarly sustain digital systems while their contributors can become invisible. This changes what governance owes people: dignity must extend to those making AI possible, as well as those consuming its products.

04Embodiment & consciousness

Partial and extraordinary at once

Eisenstein finds AI astonishingly capable while asking what its digital inheritance leaves out: unrecorded knowledge, bodily feeling and lived experience. He holds partial intelligence and super-intelligence together as a paradox, then questions whether consciousness belongs to a separable self at all. The essay explores that tension; it does not settle machine consciousness through the dialogue it reproduces.

05Science & learnability

Can nature teach us what is learnable?

Hassabis treats AI as a way to investigate why apparently immense natural problems can become tractable. Evolution and repeated physical processes may leave structure that learning systems can discover. He presents this as a conjecture, illustrated by protein folding, and distinguishes natural patterns from abstract problems where useful structure may be absent.

06Intelligence & limits

What remains slow when intelligence is abundant?

Amodei asks what limits progress once intelligence becomes cheap and plentiful. Experiments still take time, data can be missing, and institutions and physical laws still matter. His optimistic account therefore turns on a changing relationship between intelligence and its bottlenecks: AI may work around some constraints, while others remain binding.

07Ownership & public life

Who owns the ground beneath AI?

Gurumurthy treats data and computing infrastructure as foundations of a society's ability to innovate. When companies enclose collectively produced knowledge, communities can lose both economic value and the power to define useful AI. She calls for public and community institutions to govern that foundation democratically, while retaining a role for private enterprise inside an ecosystem directed toward public purposes.

08Work, responsibility and power

Who serves whom in the human–AI team?

Doctorow distinguishes a person directing a machine from a machine directing a person. A nominal human overseer can become the part that absorbs blame, while an employer captures the savings from automation. The important question is therefore how work, responsibility and power are arranged, not simply whether a human remains somewhere in the process.

09Work & political power

A share of wealth, or power to claim it?

Brynjolfsson's Turing Trap distinguishes a richer economy from one in which people can insist on sharing its gains. If machines primarily replace workers, economic dependence can become political dependence. AI that expands people's capabilities can preserve their leverage. His warning is about who can shape the bargain, not simply how large the economic pie becomes.

10Ecology and demand

Efficiency can make the footprint grow

In work with Emma Strubell and Kate Crawford, Luccioni examines a paradox: making each AI operation cheaper and more efficient can encourage enough additional use to increase total resource demand. The question becomes how AI is deployed and what demand it creates, not simply how little energy one model needs. Environmental judgment must include these wider effects.

11Awe & human possibility

Hope without a promised ending

Silva describes awe as the moment technological possibility outgrows our inherited picture of the world. He revisits his earlier hope that accelerating technology could answer his fear of death, now distinguishing an open future from guaranteed rescue. AI expands agency without deciding its purpose; wonder therefore needs judgment, compassion and responsibility alongside it.

12Relationships and collective agency

Data as soil, not oil

Tang contrasts extracting a person's data with tending the relationships in a community. An AI placed inside a group conversation must attend to coordination among people, rather than merely please one subscriber. Locally steered systems could become temporary, bounded helpers whose value lies in what they enable people to do together, and which can fade when no longer needed.

13Safety & control

The danger is the safety response

“The slogan of the Antichrist is 'peace and safety.' And we've submitted to it.”

Thiel inverts the existential-risk frame. The Antichrist of the old stories was an evil scientist with a doomsday machine; in his telling the figure who takes the world is the one who talks about Armageddon nonstop and offers peace and safety in exchange for control. Nuclear weapons, climate, bioweapons and AI each create the demand for a one-world regulator, and that regulator, not the machine, is the catastrophe he plans around.