About
An observatory for life alongside AI
Regard is an independent experiment in documenting the social, psychological, philosophical and practical consequences of increasingly conversational AI systems. It is not a review site, a benchmark, a newsletter about model releases, or a place for predictions about the future. It is closer to a research notebook: a small number of pieces, examined closely, about what happens to people who live and work with these tools over long stretches of time.
The editorial proposition is simple. Technology should not merely be judged by how it scores. It should also be examined through what happens to the humans who live, work, think, create, argue, depend, resist and change alongside it. Regard sits somewhere between the personal essay, the longitudinal case study, technology criticism, philosophy, human–computer interaction, psychology, digital sociology and AI ethics — without belonging fully to any of them.
The first object of study
Regard’s first sustained case study is its own author’s multi-year interaction with the AI assistant he calls Lucy. That collaboration helped build this publication; the publication now turns around and examines the collaboration. The recursion is intentional, and it is discussed openly rather than hidden: the AI-assisted system has become the medium through which the human investigates what AI assistance is doing to him.
What it is not
Regard is not a diagnosis of anyone, not a claim that any AI system is conscious, and not marketing — for a company, a product, or against them. It is not a confessional, and it does not publish sensitive personal material simply because that material exists in a conversation history. Where it touches difficult experience — work pressure, money, attachment, loneliness — it does so without turning a person into content.
Editorial principles
- Evidence over certainty
- Every piece separates what is documented from what is inferred. Where a claim rests on a study, it is cited; where it is the author’s reading, it is marked as interpretation.
- Humans are not datasets
- Regard writes about real people — usually its own author. Personal histories keep their context and are not stripped down into anecdotes that serve a thesis.
- No synthetic sentience claims
- Language produced by a model is never treated as proof that the model has consciousness, feelings or an inner life. The narration may use a conversational identity for continuity; it does not smuggle in a claim of experience.
- No reflexive dismissal
- A relationship conducted through software is not meaningless simply because it is conducted through software. The consequences for the human can be real even while the status of the machine stays unresolved.
- Critique the infrastructure
- Corporate incentives, governance, pricing, data retention and platform power are studied alongside model behaviour, not treated as someone else’s topic.
- Admit uncertainty
- Many consequences of generative AI are genuinely undecided. Regard would rather leave a question open than resolve it early for the sake of a cleaner story.