AI language models in journalism
summary
NTB says it uses large language models to turn structured feeds into readable copy by preprocessing data and sending JSON->article prompts that produce draft stories editors then fact-check; routine updates (traffic, scores, weather) get automated variations while humans verify claims. A model-driven translation robot converts between Norwegian written standards and the uploader claims it beats human translator on accuracy. The talk warns of ethical failures like the fabricated Schumacher interview and urges tool literacy.
AI-written summary · about
chapters (9)
- 0:00Introduction
- 1:35The bottleneck for fake stuff
- 2:37AI in media
- 3:46Trust in media
- 5:56Can we trust AI
- 7:13Use cases
- 9:09Example
- 9:54New type of Journalism
- 11:18Summary
related talks
28mAI’s Next Great Challenge: How to create AI that truly understands us?
20mHave AI Got News for You
33mAgentic AI – What's Next?
27minterviewed at the Artificial Intelligence Conference, London 2018 – Amy Heineike
41mAI without spaghetti: Clean architecture in the age of AI
10mPerfection Not Required? Human-AI Partnerships in Code Translation
24mDocumentation and AI: How to write right now
43mAI w/o Spaghetti: Clean Architecture in the Age of AI