AI policy.
Software helps with the preparation. It publishes nothing. Responsibility for every sentence rests with people.
What we use AI for
- sorting and summarising our own research material
- structured extraction of fields from publications, checked afterwards against the original
- suggested translation variants, which a specialist then reviews
- spotting contradictions and gaps in a draft
- internal drafts and outlines
- technical checks: dead links, missing image descriptions, invalid structured data
What we do not use AI for
- automatic publication, in any language
- generating source references; a Digital Object Identifier is always verified in the database
- grading evidence or transferability
- inventing figures, names, institutes or study results
- generating images that could appear to document a real case
Illustrations
Some of the illustrations on this website were produced with generative image tools. These images are atmospheric and abstracting, never documentary: they show no identifiable people, no real cases, no anatomical depictions from which anything could be inferred. Anatomical diagrams and data graphics are drawn from checked data, not generated.
Responsibility
Every published piece of content has a human body answerable for it. The use of tools changes nothing about that and mitigates nothing. We answer for what stands here as though every sentence had been written by hand, because that is exactly how it is checked.
Services used for research and editing receive no personal data about our readers.
The evidence letter
Once a month: new syntheses, checked studies, and what has changed in the state of the evidence.