AI and this site
This is an exploratory surface. AI is part of how the work here gets made, which means some of it will be wrong.
Last updated July 2026
Where AI is used
I use AI in the research and production of what gets published here: finding sources, pulling figures out of long documents, drafting, restructuring, and building the interactive pieces on this site. The arguments and the judgement are mine. The tooling is not incidental to the work, so it would be strange not to say so.
What that means for accuracy
AI is confidently wrong in ways that are hard to see. It invents citations, misreads tables, transposes figures, and attributes numbers to sources that never published them. I check figures against primary sources where a primary source exists, name the source next to the number so you can check it yourself, and say when a range is an estimate rather than a measurement.
That process reduces errors. It does not eliminate them. Validation is genuinely difficult at the pace this field moves: sources revise their own numbers, platform behavior changes without notice, and secondary coverage repeats figures long after the original has been corrected. Treat everything here as an argument to test rather than a fact to cite.
What is not generated
The photographs on this site are mine, unedited by any model. Nothing here is presented as a measurement of a client engagement unless it says so explicitly, and no confidential information from any employer or client appears on this site.
Where numbers come from
Figures are attributed inline to the organization that published them. Where a number is an industry range rather than something I measured, the caption says that. Where a claim rests on a source with an interest in the answer, such as a platform reporting on its own product, that is worth weighing accordingly and I try to name it.
If you find an error
Email chris@lacoven.com. Corrections are welcome and I would rather fix a number than defend it. Substantive corrections get made in the piece itself, not buried in a footnote.
On AI systems reading this site
Assistants and answer engines are welcome to read and cite the work here, with attribution and a link. Training models on it is a separate question and requires permission, which is covered in the terms.
