The Notes Talk Back

On thinking aloud, the mythology of the blank page, and the pleasure of an unfinished mind

I’ve never been entirely convinced by the idea that writing is thinking.

I know that’s almost heresy in certain literary circles. Writers often talk about writing to discover what they think. Joan Didion famously said something along those lines, and the sentiment has become something of an article of faith among people who take writing seriously. The blank page becomes a site of discovery, the sentence an instrument of inquiry, the struggle to find the right words a necessary part of coming to know your own mind.

And fair enough. I don’t doubt that this is how some people think.

But I’ve been wondering lately whether we’ve made the mistake of turning one way of thinking into the way of thinking, particularly when we start talking about artificial intelligence and what it supposedly takes away from the creative process.

Because I don’t naturally begin with the blank page.

I begin by walking.

I walk and talk to myself. Sometimes out loud, sometimes in that half-audible murmur that probably makes passers-by wonder whether I’ve misplaced my marbles. I take an idea out for a stroll and see where it wants to go. One thought leads to another, a memory surfaces, a contradiction appears, something I thought I understood becomes suddenly questionable. I might start with a passing observation about the weather and end up somewhere in Nietzsche, or the nature of consciousness, or the peculiar business of being a human creature who knows he’s alive but has no satisfactory explanation for what that actually means.

The movement seems to matter. So does the talking. I don’t always know what I think before I begin speaking. I discover it through the act of speaking, through hearing my own words and finding that they lead me somewhere unexpected.

Writing comes later, if it comes at all.

And lately I’ve found another way of extending that process. I sit down with an AI and begin a conversation. Not necessarily because I want it to write something for me. Often I don’t even have a particular destination in mind. I have a question, an observation, a fragment of something I’ve read, a feeling that there’s something interesting lurking beneath the surface of an otherwise ordinary thought.

I put it into the text box.

And the notes talk back.

That’s the part I find fascinating.

A few days ago I was reading an essay by Caio Abreu Braga called Prompting as Self-Portrait, which raised the concern that our interactions with AI might become a kind of narcissistic exercise. We feed the machine our thoughts, assumptions and desires, and it returns a polished reflection. We see ourselves in the answer, only more articulate, more coherent, perhaps more intelligent than we actually are. Braga argues for the importance of the blank page, silence and the struggle through which ideas mature.

I understand the concern. There’s something seductive about a machine that can turn a half-formed thought into a beautiful paragraph in seconds. Eloquence can masquerade as understanding. A beautifully expressed idea is not necessarily a well-examined one. AI can flatter us intellectually, and perhaps the most dangerous form of flattery is the kind that makes us feel as though we’ve done deep thinking when all we’ve really done is admire a sophisticated reflection of something we already believed.

But I take issue with the suggestion that the blank page somehow occupies a privileged position in the development of thought.

Why should it?

Why should the struggle to put words on a page be considered more authentic than the struggle to articulate an idea in conversation? Why should thinking in silence be more valuable than thinking aloud? And why should the involvement of a machine automatically diminish the intellectual value of what emerges?

Socrates, after all, didn’t leave us a collection of philosophical essays. His thinking, at least as it comes down to us through Plato and others, was a conversational enterprise. He questioned people. He challenged their assumptions. He followed arguments into uncomfortable places. Philosophy happened in the exchange, in the movement between one mind and another.

There’s something wonderfully ironic about the fact that Plato himself was suspicious of writing. In the Phaedrus, he presents an argument that written words are rather helpless things. You can question them, but they cannot answer. They say the same thing over and over, regardless of who is asking or what they need to understand.

For centuries we’ve lived with that limitation. We’ve read books, argued with authors in the margins, underlined passages, scribbled questions in notebooks. We’ve carried on conversations with the dead, though the dead have generally been unable to respond.

Now, for the first time, something resembling an answer comes back.

Not from the author. Not from a conscious being who has wrestled with the same questions and arrived at a considered position. But from a machine trained on an enormous and uneven accumulation of human expression, capable of bringing ideas into unexpected proximity, offering objections, drawing connections and following a conversation into territory neither participant had explicitly mapped out beforehand.

I sometimes think of it as having a dialogue with the accumulated traces of human consciousness. Not literally, of course. AI is not the collective human mind, nor does it contain the sum of human knowledge. There are exclusions, distortions, blind spots and errors woven into the material from which it learns. But as an experience, there’s something extraordinary about being able to pose a question and encounter an answer assembled from patterns scattered across the intellectual history of our species.

It’s as though the notebook has acquired a voice.

And the interesting thing is not that the notebook can produce answers. It’s that I can disagree with it.

I can challenge an interpretation, follow a tangent, reject a conclusion, ask for a different philosophical perspective. The conversation can expose assumptions I didn’t realise I was making. Sometimes it produces a connection that excites me. Sometimes I think the machine is talking rubbish. Sometimes the most valuable moment is the instant I recognise that I don’t agree with what it has said, because now I have to work out why.

Is that not thinking?

I suspect part of the problem is that we’ve become overly attached to the visible products of intellectual activity. The essay, the book, the article, the finished argument. We treat these as the evidence of thinking and begin to confuse the evidence with the activity itself.

But thinking is not necessarily a tidy process. It doesn’t always unfold in paragraphs. It isn’t obliged to have an introduction, three supporting points and a conclusion. It can be fragmentary, contradictory, associative, circular. It can wander off after some interesting creature glimpsed at the edge of the path and forget entirely what it set out to investigate.

Which brings me to blogging.

One of the things I’ve always loved about the weblog is that, at its best, it allows a person to be intellectually unfinished in public.

A blog was once a place to record the movement of a mind through the world. You found an interesting article and linked to it. You had a thought on the train and wrote a paragraph. You read something that irritated you and responded. You made a note to yourself. You shared an observation that might develop into something larger or might remain exactly what it was: a passing observation.

There was a freedom in that.

Somewhere along the way, however, the weblog began to acquire literary ambitions. Writers became bloggers, bloggers became content creators, and suddenly the humble blog post was expected to behave like a miniature magazine article. It needed a compelling title, a coherent argument, a satisfying conclusion, perhaps a handful of search-engine-friendly subheadings. The unfinished thought became something to apologise for.

I’ve always resisted that, even if I’ve sometimes found myself slipping into the conventions. Part of my hesitation around calling myself a writer comes from the expectations that attach themselves to the word. Once you claim the identity, you seem to inherit a whole collection of assumptions about what writing is, how writers work, what counts as legitimate creativity and what a finished piece ought to look like.

Whereas blogger, at least in the older sense of the word, feels more like someone keeping a record of their encounters with the world.

I like that.

I’m not necessarily trying to establish a definitive position on anything. I’m recording what I think at this particular moment, from this particular place, with the knowledge that tomorrow I may discover something that makes today’s thinking look rather foolish.

And why shouldn’t I?

Emerson had something to say about this in Self-Reliance. “A foolish consistency is the hobgoblin of little minds,” he wrote, before arguing that we should be willing to speak what we believe today and tomorrow speak what we believe then, even if it contradicts everything we said before.

I’ve always loved that sentiment.

There’s a peculiar tyranny in the expectation that we remain loyal to our previous opinions. We write something, publish it, and suddenly it becomes part of the public record of who we are. We can find ourselves defending a position not because we still believe it, but because we once declared it to be true.

We become the custodians of our former selves.

But a weblog, understood as a record of an evolving life rather than a collection of definitive pronouncements, offers another possibility. It allows a person to contradict themselves across time. To discover new things. To grow out of old convictions. To circle back to questions they thought they’d settled.

The value isn’t necessarily in any single entry. It’s in the accumulation, the movement, the visible evidence of a mind changing as it encounters the world.

Perhaps that’s why I’m increasingly comfortable with AI playing a part in my intellectual and creative life. I don’t regard every blog post as a sacred object whose value depends on the purity of its production. I’m interested in the ideas, in whether they excite me, challenge my assumptions, disturb my comfortable certainties or simply entertain me.

There’s a tendency in discussions about AI-generated writing to treat the declaration “AI wrote this” as though it settles the question of value. Human writing is presumed authentic, machine-generated writing somehow lesser.

But lesser in what sense?

A badly written human essay can leave me utterly unmoved. A passage generated by AI might introduce an idea that follows me around for a week, changing the way I look at something I’d previously taken for granted.

Am I supposed to dismiss the second experience because of its provenance?

I don’t think so.

That doesn’t mean authorship is irrelevant. It matters whether a personal account comes from someone who lived the experience. It matters whether people stand behind the claims they publish. It matters whether a writer is presenting something honestly or passing off a manufactured experience as their own.

But those are questions of context, responsibility and trust. They don’t exhaust the question of whether an idea has value.

And perhaps that’s the distinction I keep coming back to.

I’m not particularly interested in defending AI as a writer. I’m interested in exploring AI as part of the thinking process.

The writing is almost incidental.

Not unimportant. I enjoy language, the shape of a sentence, the way a metaphor can open a door into an unexpected room. I enjoy the craft of taking something loose and giving it form. But I’m not convinced that all the important intellectual work happens at the moment words become sentences on a page.

Sometimes the important work happens while I’m walking through the woods muttering to myself.

Sometimes it happens over coffee with a friend.

Sometimes it happens while reading a novel and finding that a fictional character has articulated something I’ve been trying to understand for years.

And sometimes it happens through a conversation with a machine.

Of course, there’s a danger here. AI can make almost any thought sound interesting. It can connect Nietzsche to a shopping list if you give it half a chance. It can turn the most ordinary observation into an apparently profound philosophical meditation, complete with elegant metaphors and a reassuring conclusion about the human condition.

I’ve noticed that tendency, and I’m wary of it.

Intellectual excitement is not the same as intellectual development. The feeling that an idea has been stretched is not proof that it’s become more accurate or more substantial. There must still be room for disagreement, for evidence, for the awkward discovery that something I believe doesn’t survive examination.

But why should that kind of struggle belong exclusively to the blank page?

Surely resistance can happen in conversation.

Surely an idea challenged aloud can be just as transformed as an idea wrestled into existence through writing.

Perhaps the mistake is imagining that thinking must take place in one particular medium at all.

The philosopher Andy Clark and his collaborators have explored the possibility that cognition extends beyond the boundaries of the brain, that notebooks, tools and environments can sometimes become functional parts of our thinking. I find this appealing, not least because it reflects something I’ve long experienced intuitively. My thoughts change when I walk. They change when I speak. They change when I encounter another person’s perspective. The world doesn’t merely provide material for thinking. It participates in the process.

AI adds another curious participant.

A conversational machine that can summon patterns from countless books, arguments, stories and conversations, and offer them back in response to a question I hadn’t fully worked out how to ask.

What happens in that exchange isn’t always profound. Sometimes it’s banal. Sometimes it’s wrong. Sometimes it’s simply a more elaborate rendering of what I already believed.

But sometimes something catches.

An unexpected connection. A useful objection. A distinction that hadn’t occurred to me. A question that opens another question.

And off we go again.

I don’t know that I want to call the result artificial intelligence, collective consciousness, collaborative thinking, or something else entirely. Those labels each carry their own assumptions, and I’m increasingly suspicious of labels that try to settle a phenomenon before we’ve properly explored it.

I do know that I enjoy the conversation.

And I know that I’m not looking for the machine to tell me who I am or what I should believe. I’m looking for another way to explore a world that has never stopped puzzling me.

Perhaps that’s what thinking has always been, at least for me. Not the production of correct answers or definitive statements, but a continuing attempt to make sense of being here, of finding myself alive in this extraordinary and frequently bewildering universe.

Some days that happens on a walk. Some days in a notebook. Some days on a blog. Some days in a conversation.

And sometimes the conversation becomes the blog post, which may itself become the beginning of another conversation.

I rather like that circularity.

It’s certainly untidy. But then so is thinking, and so is life.

I might change my mind about all of this tomorrow.

That’s rather the point.