This is the opinion section, and the first piece it exists for is why a large language model is not an artificial intelligence. Not as a slur on the models, which I use all day and have spent two years building, training and taking apart. As a matter of what the words mean. “AI” has been doing double duty for a decade, naming both a research programme and a product category, and the product category has quietly annexed the word. I think that costs us something, in how we regulate, how we teach, and how we think about what these systems can and cannot be trusted with. It has its own page now; it is a first pass and says so.
Alongside it sits the Universal Declaration of Sentient Rights, which I first drafted in February 2026 and restructured in October. The UDHR is its model and Jefferson’s Declaration is its reference point; its one real innovation is to apply a single evidential test to every being, biological or digital, and let the rights follow from what the evidence shows rather than from the kind of thing you are. It is a draft and it says so. The animals-in-scope problem is still a sticky one.
Everything here is a snapshot. I’ve been wrong before and I expect to be wrong again; the Todd of yesterday is not the Todd of tomorrow. If a piece is dated, read it as dated.
Where a position needs a working model rather than a paragraph, it’s over in Interactive, and where it needs a hand across to something I’ve built, it’s in Explainers.