Jeffrey Dvorkin Explores AI and Human Judgment in Journalism

Artificial intelligence can expand a journalist’s capabilities, but the profession risks losing context, independent thought and public purpose if technology becomes a substitute for editorial judgment, veteran journalist Jeffrey Dvorkin says.

In an interview with Energi Media’s Markham Hislop, the author of Trusting the News in a Digital Age discussed how journalists can use AI while remaining responsible for their work. The conversation examined human oversight, the value of writing, ownership of reporting and the commercial pressures shaping newsrooms.

Dvorkin said his book grew from teaching journalism at the University of Toronto Scarborough. Students understood media technology, he said, but often lacked the wider knowledge needed to interpret events. Newsroom cuts compounded that problem by removing experienced journalists who could provide context.

He framed journalism’s purpose around an audience understood as “citizens first and consumers of information second.”

Three Models for AI and Journalism

Hislop proposed three models. In artisanal journalism, the reporter conducts the research and interviews, verifies information and writes the story. In news as content, AI handles much of the production, leaving humans with a smaller role.

The third model, called Bob, treats the journalist as a journeyman and AI as an apprentice. The journalist directs the work and retains final editorial responsibility.

Hislop described using an AI assistant, Clara, to search a collection of reports, datasets and interview transcripts. The process helps assemble evidence and develop an argument, while he writes the lead and conclusion and decides whether the work is satisfactory.

He also described an editorial constitution intended to guide the assistant’s work, including commitments to truth, accuracy and democratic values. The assistant can question his claims, prompting further consideration.

Writing Is Part of Thinking

Dvorkin responded positively to the collaboration model but challenged a different vision in which journalists gather facts and leave writing to AI.

He recounted a former student’s description of that approach at another institution. His objection was that writing and repeated revision clarify a journalist’s thinking. Removing that work risks diminishing the journalist’s contribution, he argued.

The discussion also explored ownership and attribution when AI systems draw on material created by journalists and news organizations. Dvorkin raised questions about who benefits from that intellectual production and how its origins should be acknowledged.

He connected those concerns to earlier newsroom practices of following another outlet’s reporting without credit. The conversation left the ownership questions open.

Who Benefits From Greater Productivity

Hislop argued that AI can improve journalism by finding evidence and analysis he would struggle to locate alone, particularly when connecting international developments to Canada.

Dvorkin agreed, while warning about management approaches that judge activity primarily by its financial return. Greater technical capability does not settle how news organizations will use it.

For Hislop, that leaves an ongoing public editorial experiment whose methods remain open to scrutiny. The interview ended with a question Dvorkin remembered from a former NPR boss: “Is there a downside?”

For journalism’s AI experiments, the practical challenge is to preserve the judgment, scrutiny and responsibility that make reporting useful to the public. Both speakers supported continuing that examination through open discussion of practice.