Who's Shaping Who?

John Nosta — "When AI Changes the Shape of Thought"

Nosta's claim is that AI doesn't just answer our questions, it exerts a kind of 'cognitive gravity' — a steady pull that bends how we think. We borrow its pace, mistake its fluent confidence for real understanding, and hand over the formative early stages of a thought, until we're shaping ideas for an algorithm's ear rather than a person's. And the bend may not be temporary: 'once AI bends a thought,' he warns, 'it rarely bends back.' The most vivid short read on the list for the evening's core worry — how is it changing the way we think?

Thomas Germain — "How to Stop AI From Turning Your Brain to Mush"

Germain's BBC Future piece starts from a worry with precedent: GPS dulled our sense of direction and search engines thinned our memory, and researchers now fear AI could do the same to creativity and critical thinking — a muscle left to waste. But it lands as a practical guide rather than a lament: how to hand the grunt work to these tools without also handing over the thinking that keeps you sharp — how to 'think outside the bots.' The most do-this-instead answer on the list to the evening's worry about what we're quietly giving up.

Adam Green — "AI Is Creative. But Can It Replace Human Creativity?"

Green, a Georgetown neuroscientist, put the question to a real test: analysing more than 370,000 college-admissions essays written before and after ChatGPT, he found the AI-assisted ones reached for richer vocabulary yet carried fewer genuinely original ideas. His worry isn't that machines can't be creative — it's that leaning on them costs us the practice of making something new ourselves, the very muscle hard problems demand. 'If we lose the diversity of new ideas ... we are not likely to solve those big societal or global problems,' he warns. A rare data-backed answer to the evening's question — and its verdict lands on 'maybe only if we let it.'

Nicholas Carr — "Is Google Making Us Stupid?"

Written in 2008 but sharper than ever, Carr's Atlantic essay asks what happens to human cognition when we outsource memory and attention to machines — how frictionless retrieval of everything may be quietly reshaping our capacity for deep, sustained thought. It predates the chatbots but frames the exact worry AI intensifies, and names the evening's central anxiety directly: which capacities thin when we stop using them?

Giles McMullen — "The Cognitive Science of Using AI"

A clear walkthrough of the MIT Media Lab study 'Your Brain on ChatGPT,' which put people under EEG while they wrote essays with and without AI. The AI-assisted writers showed markedly lower brain connectivity, remembered little of what they'd 'written,' and felt scant ownership of it — the study that put the phrase 'cognitive debt' into the conversation. The evening's worry, turned into something you can watch happen on a brain scan.

Imran Gardezi — "Cognitive Debt"

Gardezi takes the idea of cognitive debt out of the lab and onto a working software team: when AI writes production code that runs but no one can fully explain, velocity climbs while understanding quietly drains away. He argues this isn't technical debt but something subtler — a shared comprehension that erodes until, months later, a single break forces the whole system to be relearned from scratch. A concrete, real-stakes look at what it costs to let the machine do the thinking — and three safeguards for keeping it a tool.

Brian Christian — The Alignment Problem

Christian's lucid, reported tour of the field's central technical-and-moral question: how do you get a system that learns from data to actually do what we mean — and whose values does it encode when it does? He braids the history of machine learning together with the ethical failures already visible (biased models, gamed objectives) and the researchers trying to fix them. The best single book for understanding how these systems really work without a maths degree.

Nicholas Carr — The Shallows

The book-length case behind Carr's famous essay, and a Pulitzer finalist: the internet, he argues, isn't just changing what we read but rewiring how we read — training the brain for restless skimming while the older capacity for deep, sustained attention quietly atrophies. Drawing on neuroscience and the history of earlier mind-altering technologies — the map, the mechanical clock, the printed page — he shows that our tools reshape our thinking whether we notice or not. The deep-end argument for the very worry the evening keeps circling: what happens to the mind that stops diving.

C.S. Lewis — The Abolition of Man

A short, dense 1943 classic that has become strangely urgent. Lewis argues that 'man's conquest of nature' is really some men gaining power over others, and that a technological age which discards a shared vision of the good will end up producing people it can shape at will — 'men without chests.' The philosophical spine of the evening: before asking what machines can do, it asks what a human being is for.

UC Davis Health — a voice, decoded from the brain

A man with ALS who had lost the ability to speak has it back — not through his mouth but through a brain-computer interface that reads the neural signals for the words he intends and, with AI, voices them aloud at up to 97% accuracy, in something close to his own former voice. It is among the most accurate systems of its kind ever built. AI not as a replacement for the person, but as the bridge that carries his own words back into the room.

Wandercraft — a self-balancing exoskeleton

Wandercraft's self-balancing, AI-stabilised exoskeleton lets people with walking disabilities stand and walk without crutches or a walker — the device handling balance itself, hands free, so the person can simply move. Built first for hospital rehab and now heading toward a personal version for the home, it turns 'assistive technology' into something much closer to walking. The hope made literally load-bearing: the machine gives the body back to the person.

Google — a guide for blind runners

Google's Running Agent lets blind and low-vision runners run on their own — no tethered guide, no rope, no volunteer — using a phone camera and on-device AI to follow the lane line and steer the runner with sound. What was impossible without another person's arm becomes something you can do alone, at your own pace. AI as independence: not running for you, but clearing the way so you can.

Penda Health — a second opinion in every exam room

At Penda Health's clinics in Kenya, an AI tool called AI Consult reads along with the clinician in real time, flagging a possible missed diagnosis or an unsafe prescription before the patient leaves the room — and in study it meaningfully cut clinical errors. It is a quiet safety net that makes ordinary primary care safer without removing the human clinician. AI as a second set of eyes for the places with the fewest specialists, not a substitute for the doctor in the room.

Museum of Jewish Heritage — conversations with survivors, kept alive

Visitors can sit before a Holocaust survivor and ask their own questions — how did you survive, what did you lose, how do you forgive — and hear the survivor answer, drawn in real time from hours of filmed interview by AI that matches each question to the right recorded response. As the last survivors pass, it keeps the encounter possible: not a chatbot imitating the dead, but a way to keep asking real people, in their own words and faces, what they lived. AI in the service of memory — so testimony can still answer back.