AI policy

We use AI tools to help us work faster. We never let them decide what is true.

Where AI helps

• Finding candidate research papers and extracting details from abstracts.
• Drafting and editing copy, and suggesting internal links.
• Checking that every number on a page matches its source abstract.

Where AI is never allowed to decide

• Evidence grades. AI may propose a grade with reasons; no grade is final until it has been validated by a human editor. Until then, pages show the grade as provisional.
• References. No reference is published unless it has been verified against its PubMed record.
• Safety information. Helpline and emergency details come only from a controlled list checked against each organisation’s own website — never from AI.
• People. We never invent experts, reviewers, qualifications or quotes.

Ask dopamine.co.uk is not a chatbot

Ask matches your question to pages we have already researched and published, and shows those pages’ own summaries. It does not generate text, so it cannot make up a medical answer. If we haven’t covered something, it says so.

No AI conversations with users

We do not offer an AI chat. “Talk it through” is a set of pre-written questions. Any future conversational feature would only launch after the safeguards in our clinical safety statement.