Everybody’s Smart · Interview series
In the making

What does thinking with AI do to the thought?

A series of conversations with researchers in cognition, phenomenology and the philosophy of mind — exploring what happens to human knowing, imagining and judging when a reasoning partner enters the loop.

Long-form + newsletter Researcher conversations Cartographic, not argumentative
AI and Judgment Creative Labor and Authorship Everybody’s Smart
Series dossierEverybody’s Smart
The spine question What does thinking with AI do to the thought?
The cut Not engineering, not ethics — the phenomenological and epistemological question of what happens to the cognitive act itself.
Who’s in the room Researchers in cognition, phenomenology and the philosophy of mind — people who study knowing from the inside.
Where it lives A long-form record here, condensed for the Everybody’s Smart newsletter. Cartographic, not argumentative.
01 — Purpose

The premise

Work was assigned before we had a reasoning partner. The division of cognitive labor — what belongs to human judgment, what belongs to calculation — was settled long before AI could reason. Now that settlement is dissolving, and no one has renegotiated the terms.

This series is not about whether AI is good or bad for thought. It is not about productivity, accuracy, or bias. It is about something more fundamental: what happens to the cognitive act itself — to knowing, imagining, understanding — when an entity that reasons enters the process alongside you?

The researchers I’m looking for are not building AI systems. They are not auditing them. They study how knowing happens from the inside — the texture, the conditions, the fragility of human understanding. People who would recognize this as a genuinely new problem, not a variant of an old one.

The boundary between your mind and the tool is not where you think it is — and it moves depending on what cognitive act you’re performing.

02 — Topic

The intellectual territory

The series maps two distinct cognitive landscapes that AI reshapes in different ways. These are not stages of a single process — they are different orientations toward the unknown.

The epistemic map — when you’re trying to know
Zone 1
Augmented recall

AI surfaces what you already knew but hadn’t activated. Phenomenologically seamless — it feels like thinking. Epistemologically safe.

Zone 2
Legitimate expansion

AI introduces adjacent territory you can evaluate and integrate. Your existing knowledge acts as a validation filter. You recognize it as true for you.

Zone 3 Needs a name
Context drift

AI introduces knowledge accurate in itself but calibrated to a different situation. Not wrong, not hallucinated — someone else’s truth in the wrong frame. The knower cannot detect the misalignment.

Zone 4
Epistemic laundering

Unmetabolized knowledge performing as owned. The person believes they know it, but they haven’t done the cognitive work that converts information into situated understanding. It spreads before it fails.

The imaginative map — when you’re trying to see differently

When seeking inspiration rather than knowledge, the cognitive stance inverts. You are not filling a frame — you are suspending it. AI’s tendency to generate prolifically and without self-censorship becomes a different kind of resource: an imagination prosthetic that extends the range of the conceivable before judgment enters.

The questions here are not about accuracy. They are about whether surprise was genuinely generative or merely novel-feeling; whether recognition is close enough to generation to matter; and whether using such a prosthetic develops, atrophies, or reshapes what you are capable of imagining on your own.

03 — The spine

One question, many angles

What does thinking with AI do to the thought?

This is the question every conversation answers from a different angle. It admits no clean resolution. It is epistemological before it is technical. And each researcher brings to it a conceptual vocabulary that builds something cumulative across episodes.

04 — Who I’m talking to

The kinds of minds

Five overlapping profiles. People who’d answer the spine question from genuinely different directions, building something that compounds across episodes.

Post-phenomenologistsResearchers examining how technologies mediate human experience and cognition — the conditions under which a tool becomes part of the perceptual act. (Ihde / Verbeek lineage.)
Situated & embodied cognition researchersWorking in the tradition of Dreyfus, Hutchins and enactivism — cognition as inseparable from context, body and environment.
Philosophers of expertiseExamining what skilled judgment consists of, and what distinguishes it from knowledge that is merely held.
Epistemologists of testimony & social knowledgeHow knowledge travels between minds — what is lost or transformed in transmission, and what conditions genuine understanding.
Cognitive scientists studying metacognitionKnowing what you know — and the conditions under which that self-knowledge fails.
05 — References & inspirations

The lineages it draws on

The series doesn’t start from zero. It picks up threads from traditions that have been studying the mind–tool boundary for decades — long before generative AI made the question urgent.

PostphenomenologyDon Ihde’s analysis of human–technology relations and Peter-Paul Verbeek’s technological mediation — tools don’t just assist, they reshape the acts they enter.
Skilled coping & situated cognitionHubert Dreyfus on expertise beyond rules, and Edwin Hutchins on cognition distributed across people, tools and environments.
4E cognition & the extended mindEmbodied, embedded, extended and enactive accounts that refuse to locate thinking strictly inside the skull.
Social epistemology & testimonyHow belief travels between knowers, and what conditions warrant trusting what we’re told — now that the teller can reason.
Metacognition researchThe study of knowing what you know, and the systematic ways that self-knowledge breaks down.
06 — Format

How a conversation happens

Participants choose their preferred mode. The range is deliberate: it accommodates time-scarce researchers while also working as a provocation aligned with the series subject.

Option A
Written, by email

A set of questions answered at your own pace, in your own register.

Option B
Async AI-assisted dialogue

You receive a Claude link; AI asks follow-up questions you may answer or skip. A somewhat recursive arrangement.

Option C
Video conversation

A recorded conversation, edited for publication.

Each episode publishes in two forms: a long-form piece here on sergiotavares.com (the durable record), and a condensed version for the Everybody’s Smart newsletter. Episodes accumulate into a body of work that maps the terrain — not prescriptively, but cartographically.

07 — A taste

A sample question arc

  • When you work through a problem with AI, at what point — if any — does the resulting understanding feel like yours? What marks that moment?
  • Is there a form of knowing that requires having been confused first? What happens to that confusion when AI resolves it before you’ve fully inhabited it?
  • What’s the difference between being reminded of something you knew and being told something you didn’t — and can you still tell the difference when the interlocutor reasons?
  • You describe a problem to AI using the wrong frame — and AI, faithfully, answers the problem you described. What kind of error is that, and whose is it?
  • When an AI generates an image or connection that genuinely surprises you — was that your imagination, extended? Or something that arrived from outside and you recognized?
  • What in your field cannot be understood by someone who has nothing at stake in the question?

Recognize this problem?

The series is cartographic rather than argumentative — each conversation adds a layer the next one couldn’t have been written without. If you study how knowing happens from the inside, or you just want to follow where the map goes, there’s a way in.

Get in touch Read Everybody’s Smart

Part of Everybody’s Smart — culture, cognition and creativity, now that everybody’s smart. Episodes will appear here as they publish. By Sérgio Tavares.