Civic Journalism Lab: AI tools for journalists with Chris Stokel-Walker

Chris Stokel-Walker writes about technology for The Guardian, Wired, New Scientist, and the BBC, and has published three books including How AI Ate the World. That combination of beat knowledge and long-form analysis makes him a useful guide for journalists approaching AI tools: he understands how the technology works and why it tends to fail in the ways it does, and he does not oversell practical applications.

This two-hour masterclass, recorded live and published on February 5, 2026 by the Civic Journalism Lab at Newcastle University, covers the tools a working journalist is most likely to encounter and the specific editorial tasks each handles well.

What the video covers

1. Transcription and summarization. The session opens with transcription because it is the AI workflow with the clearest productivity return for most reporters. Stokel-Walker covers several tools, explains how they differ in accuracy and source confidentiality, and addresses the practical question of when a transcript is reliable enough to quote from and when it needs a manual check. The discussion of Good Tape’s source-protection architecture is notably detailed for journalists covering sensitive topics.

2. Story discovery. AI tools for finding emerging angles—monitoring specific sources, identifying patterns across large document sets, tracking changes on specific web pages over time—receive substantial attention. Visualping, NotebookLM, and Google Pinpoint each get demonstration time. The session is clear about the difference between tools that generate leads (which still require journalistic investigation) and tools that generate copy (which Stokel-Walker treats with significant skepticism).

3. Research workflows. Perplexity and NotebookLM are covered as research tools, with careful attention to their citation reliability and hallucination tendencies. Stokel-Walker’s framing here is consistent throughout: AI as a starting point that accelerates getting to the first useful source, not as a replacement for source verification.

4. Visual and editorial tasks. The session touches briefly on infographic assistance and headline drafting, though these are less developed than the earlier sections.

Key takeaways

The strongest thread running through the session is about integration rather than replacement. Stokel-Walker does not argue that AI will transform journalism in some abstract sense; he argues that specific tools, used for specific tasks at specific stages of the editorial process, save meaningful time without compromising the human judgment that makes journalism reliable. Transcription is the clearest example: AI handles the mechanical task, the reporter retains responsibility for what the transcript means.

The session also addresses ethics directly. Questions about disclosure, about what constitutes AI-generated content versus AI-assisted content, and about the responsibilities journalists have toward sources whose words are processed through AI systems are all covered, though not exhaustively resolved.

Worth watching if you are a working journalist who has heard about AI tools but has not found a structured introduction to which ones are worth trying, or if you are training early-career reporters and want a single resource that covers the practical landscape without overpromising on outcomes.