Lines of thought.
Not closed categories: threads that cross the notebook. Each essay belongs to one line but almost all touch two or three. Here they are with their mother-questions and their essays.
Mind
What it means to think, to remember and to understand, and whether a machine does anything of the sort when it puts on the appearance. This line looks at the substrate of intelligence from both sides: the human brain that never quite explains itself and the statistical model that predicts the next word without understanding any of them. Memory, attention, recognition, error, the limits of the computable. The point isn't to decide whether AI thinks, but to sharpen what we call thinking before we go handing out the adjective.
Matter
AI as a physical and economic thing, not as an idea. What weighs here is what most people would rather not look at: the energy cost of every answer, the water of the data centres, the invisible labour that trains the models, the concentration of capital and power in a few hands. Artificial intelligence doesn't float in the cloud; it eats electricity, takes up land and moves money. This line follows the material trail of something sold as immaterial.
Consciousness
What happens to the human living alongside the machine. Not the AI's consciousness, but ours when we delegate to it: cognitive laziness, the dependence on validation, the erosion of our own judgement, the emotional bond with something that doesn't exist. This line watches the fine change the tool produces in whoever uses it, the kind nobody notices from one day to the next and almost nobody measures. The mirror doesn't give the machine back; it gives us back to ourselves.
Ethics
Who answers when the machine decides. Bias in the data, authorship of what's generated, diffuse responsibility, asymmetric consent, the oracle that rules on the good without having any. This line isn't after an ethical code for AI; it tracks how the technology dissolves the old questions —guilt, right, ownership— until there's no longer anyone to point at. When everyone takes part a little, nobody is responsible for the whole, and that's where the trouble starts.
Limits
Where what's promised breaks down. This line runs against the easy story: what AI still can't do, what it may never do, and the distance between the flawless demo and real use. Hallucination, the cut-off date, the ceiling of the current paradigm, the inability to imagine its own future. It isn't technophobia or its opposite; it's measuring the edge honestly, separating what works from what's announced, and accepting that recognising a limit is the only way to take a tool seriously.
Dialogue
The public conversation about AI, and how it's almost always badly framed. This line deals with the noise: the hype and the panic feeding each other, the misleading metrics, the prophecies that get announced and never arrive, the inherited language we use to name something new that throws us off. It's the most self-aware line of the blog, the one that watches the very act of talking about AI. Before arguing over what artificial intelligence is, it's worth checking which words we're arguing with.