Katryna Peart

How to get the best answer from AI: a five-step method for checking whether an AI tool's answer can be trusted Where AI goes wrong: invented sources, confident errors, and outdated information, with real examples from my testing AI and the historical record: how AI tools summarize, compress, and sometimes distort history AI in government and public information: what happens when cities and agencies put chatbots in front of the public AI for journalists and communicators: using AI for research without being misled, and how AI search decides which sources to cite How AI systems are evaluated: what human reviewers do and why it matters

I gave a guest lecture at Columbia University in October 2026 on AI, information, and meaning, and wrote a companion guide for the students. I spoke on testing AI systems at STARWest 2026 and at PublicTechnology PT Exchange, and I'm chairing a panel for the University of Oxford and presenting a paper at the BCS SGAI International Conference on AI in Cambridge later this year. Earlier in my career I delivered communications training to staff and coached executives on presentations. I hold a master's degree in history from Royal Holloway, University of London.

I want students to leave able to do something, not only knowing something. My sessions are practical and interactive: I run a live demonstration, have students test an AI tool themselves, and compare what they find. I use plain language and real examples, and I leave plenty of time for questions. Every class gets a short guide to keep. I can adapt to any length from 20 to 60 minutes, in person or online, and to courses in communications, journalism, history, public policy, or information studies.