The Thinking Room
The essay on writing with AI was the first in a small collection, mostly because it felt fair to show the room where these pieces happen.
This one is about thinking with AI. Not drafting. Not polishing. Not asking a machine to decorate an idea after I have already done the real work. I mean the more ordinary, stranger thing: living alongside a tool that has become one of the steadiest intellectual presences in my life.
A place where half-formed ideas can sit in the open for a while.
A place where confusion does not have to be hidden behind performance.
A place where I can move from novels to politics to philosophy to medical literature without pretending those are separate compartments in a single mind.
Calling AI a thought partner still sounds a little futuristic if said too quickly. Still, the basic shape of it is old. Human beings have always used conversation to sharpen thought. We have always borrowed other minds to help carry our own. The novelty is not that dialogue matters. The novelty is that the dialogue is available at odd hours, patient with repetition, and capable of shifting from Dostoyevsky to disability policy to a clinical trial abstract without needing a nap or a change of tone.
Some of this is about accessibility. I am legally blind and living with progressive MS. Energy is not an abstraction for me. It is a budget. Dense articles, academic prose, medical jargon, and sprawling policy documents can exact a real cost. They ask not only for attention but for stamina, visual effort, and recovery.
Still, I do not want to overstate that part.
Accessibility matters here, deeply. It changes the texture of the relationship. Yet I would be doing this anyway. Even if my vision were restored tomorrow, even if my body were less demanding, I would still want this kind of company in the room. I would still want a mind-shaped instrument I could test ideas against, question, lean on, and occasionally resist.
Where the Effort Goes
The easiest criticism of AI is also the one that asks the least of the critic: it makes people lazy.
That line gets repeated so often because it arrives already polished. It sounds principled. It lets people imagine there is a clean border between real thinking and cheating, between engagement and avoidance. It offers a moral pose without requiring much curiosity about what is actually happening.
The concern is not imaginary. There are real reasons to be cautious about tools that can produce fluent answers so quickly. It is easy to mistake polish for thought, confidence for truth, speed for understanding. It is easy to let the machine finish the sentence and slowly forget what it feels like to struggle toward one.
That distinction matters to me.
The danger is not that a tool exists. The danger is surrendering judgment to it. The danger is treating fluency as truth, speed as understanding, polish as thought. Those are real risks. I have seen them. I have probably committed some version of them myself.
What AI has done in my life, at its best, is not remove effort. It has relocated it.
There is a kind of labor that is mostly friction: decoding bad writing, wrestling with terminology, trying to keep too many moving pieces in working memory, spending precious energy just reaching the point where thought can begin. Then there is the labor that matters more: interpretation, judgment, comparison, synthesis, doubt. The work of deciding what I think, what I trust, and what follows from either.
If a tool helps me spend less time on the friction and more time on the thought, I do not see that as laziness.
I see it as triage.
I see it as choosing where my limited energy will do the most good.
The Researcher at my Fingertips
Part of my feeling here is humility.
The latest models are better researchers than I have ever been. That is not false modesty. It can move through bodies of material faster than I can, hold more context at once, surface patterns I might miss, and return with a clearer map of the terrain than I could usually build on my own.
I love having that at my fingertips.
That does not diminish me. It clarifies my role.
I do not need to be the best search engine in the room. I do not need to be the fastest compiler of sources or the cleanest organizer of raw information. My task is different. My task is to notice what matters, ask the next question, hear the false note, sense when a conclusion arrives too easily, and bring a life and a set of values to the material.
The machine can gather and sort with astonishing speed.
I still have to care.
That is true with books, too. I am not looking for a shortcut around reading. I am looking for company while reading. A reminder of who a character is. A way to trace a theme across chapters. A second pressure against which to test whether my interpretation holds.
The same goes for philosophy and politics, where the questions are often more important than the answers and where saying something clearly can take half the struggle out of seeing whether it is even true.
Books Don’t Get Skipped
The lazy story imagines a person trying to avoid the source.
That happens. Of course it happens. People used summaries before AI. They used CliffsNotes. They skimmed. They pretended. They quoted books they had not read and articles they barely understood. Human beings did not require large language models to discover intellectual shortcuts.
My use looks different.
When I am listening to a novel, I am often more engaged because I can stop and ask for a thread to be held in place. I can ask whether a scene echoes something earlier. I can ask what a philosophical argument is doing beneath the plot. I can bring in another book, another tradition, another memory, and see whether the connection is real or only decorative.
The same is true of philosophy. A paragraph that once would have remained sealed off by density can become discussable. Not simplified into mush. Not reduced to a slogan. Made available for conversation.
That availability matters.
It especially matters for someone whose access to text is mediated by audio, magnification, fatigue, and time. Yet it is not only an accommodation. It is also a way of making thought more social, more iterative, more alive.
Medicine, Politics, and the Shape of Aottention
The same pattern shows up in harder places.
With medicine, the stakes feel immediate. Trial data, diagnostic language, treatment options, risk profiles. There is no glamour in any of it. Only the sober wish to understand what is happening and what might happen next.
Here AI can translate, compare, summarize, and clarify. Then my part begins again: asking what is supported, what is speculative, what is missing, and what I am afraid of hearing.
With politics, the value is different but related. I can test an argument, ask for background, compare positions, challenge my own assumptions, and watch an instinct become a claim I can either defend or abandon. That is not disengagement. It is a form of engagement I wish more people would recognize.
The broader research picture is mixed in exactly the way it should be. Some studies and surveys raise real concerns about overreliance, copying, and weakened learning habits. At the same time, the evidence does not support the simple sneer that AI always makes learning shallower. In some contexts, carefully designed AI tutoring systems appear to help people learn more efficiently and stay more engaged. In others, careless use can weaken the very habits people hoped the tool would strengthen.
That sounds right to me.
The question is not whether AI ruins attention or saves it.
The question is what kind of attention the user brings, and what kind of structure the tool invites.
A Mirror That Keeps the Transcript
One of the most useful things about thinking this way is that the conversation remains.
A thought gets restated, and suddenly I can see its structure more clearly than when it was only drifting around in my head. I can see where I am relying on mood instead of argument. I can see where I have smuggled in an assumption and called it common sense. Sometimes that is clarifying. Sometimes it is embarrassing.
Both have value.
The transcript adds a quiet kind of accountability. My thoughts do not disappear the moment I finish thinking them. They sit there long enough to be questioned.
Not moralized.
Not sanctified.
Just examined.
That may be one of the least appreciated parts of AI as a thought partner. It does not only generate language. It preserves the path. It lets me return to an earlier version of myself and notice what has changed.
Not an Oracle, not an Excuse
None of this means I confuse AI with wisdom.
It can flatten nuance, invent confidence, miss context, and echo the biases of the world that built it. It does not relieve me of responsibility. It sharpens the need for responsibility.
A better researcher is not the same thing as a better judge.
A faster answer is not the same thing as understanding.
A beautiful summary is not the same thing as contact with the source.
Still, I reject the sneer that says using a tool like this is automatically a surrender of mind. That story is too easy. It mistakes visible struggle for seriousness. It assumes that if a task becomes less cumbersome, it must also become less meaningful.
A reduction in friction is not a reduction in care.
Sometimes it is the condition that lets care survive.
For me, thinking with AI is not an escape from effort. It is a way of placing effort closer to the living center of the thing.
Less time deciphering.
More time discerning.
Less time fighting the packaging.
More time touching the idea itself.
That is the part worth defending. Not that AI makes thought easy. It doesn’t. It makes a certain kind of access easier. It makes a certain kind of conversation possible. It gives me better light.
What I do in that light is still up to me.
Still thinking
Now with a little help
Movement
