Madison journalists weigh AI’s benefits and risks at community forum

# What happened On Sept. 17, Isthmus hosted a community discussion at Cargo Coffee East about artificial intelligence in journalism. The panel included Isthmus editor and president Judith Davidoff, Omar Waheed of Madison365, and Mike Wagner of UW–Madison's School of Journalism and Mass Communication. Angela Russell moderated.

# Why the timing mattered The conversation followed national headlines generated by Jacob Coxen, a former OpenAI and Anthropic staffer, who warned publicly about the risks of rapidly advancing AI. Panelists narrowed the focus to the local P(doom) for journalism — that is, how AI might damage local newsgathering and reporting practices.

# The local incident that sparked the talk

# How Isthmus responded Isthmus's policy permits limited AI use — for example, transcription software — but prohibits using AI to write or edit articles or to generate images. Davidoff said the newsroom has not experimented with AI drafting and that she wants to avoid using the technology in ways that sacrifice quality. She warned the worst outcome would be newsrooms demanding higher output at the expense of reporting standards.

# Panel perspectives Mike Wagner framed AI as another media-era shift that can change business models and staffing. He cautioned against accepting the narrative that AI's arrival is inevitable in a way that benefits a small group of wealthy actors while reducing newsroom capacity. He argued that treating AI as a cost-cutting productivity tool could lead publishers to hire fewer journalists and produce more homogeneous, lower-quality content.

# Practical takeaways for newsrooms

• Use AI tools selectively and document when and how they're applied. Transcription is a commonly accepted use case.
• Train reporters and editors on tools provided by their organizations, including data-handling settings and limits.
• Maintain strong editorial oversight, especially for reporters working outside familiar beats.

# What the panel didn't resolve The discussion did not advocate a single path forward. Panelists agreed on caution: AI can help with certain tasks but becomes harmful if it enables staffing cuts or substitutes for editorial judgment.

# Bottom line