Journalism schools integrate AI at the assignment level while maintaining core reporting skills

Journalism programs at three of the top public J-schools in the U.S. are voluntarily introducing AI assignments while delaying them until students have established core reporting skills, faculty said. The approach reflects a careful balance: employers want graduates who understand what AI tools can do, but the industry also demands the interviewing, fact-checking, and writing abilities that machines cannot replicate.

At the University of Kansas, the University of Oregon and the University of Missouri, instructors have not received top-down mandates to teach AI. Instead, faculty have the freedom to decide when and how to bring AI into their courses. "Your contribution is the original reporting," said Liz Lucas, journalism instructor and AI faculty fellow at the University of Missouri. "You talked to somebody, or you found this data set, or you were at this event."

Core skills and AI literacy

Several J-school faculty members explained that original reporting requires a core skill set: developing relationships with sources, asking good questions, scrutinizing and fact-checking sources, and constructing an article. At the same time, AI literacy is becoming non-negotiable. "I don't think an AI can interview a person the way the human can," Lucas said. "There are many aspects of human intelligence that are not verbal, and as of right now, that's a unique skill that we have."

Samuel Muzhingi, a Ph.D. student and teaching assistant at the University of Kansas School of Journalism and Mass Communications, is part of a research group interviewing employers about what they want from graduates. "They don't necessarily need students to be AI experts, but they do want them to understand what these tools can do," he said. "How are they useful, and maybe where can they make mistakes?"

Lucas pointed to data journalism as a bright spot, citing CNN's recent use of AI to scan 100,000 photos released by the Department of Justice and find redaction errors. "Before, you had to have some pretty advanced knowledge to be able to do that," she said. "Now it's much more accessible."

A breakdown on IBM's website notes that accepted uses of machine learning in news include interview transcription and social media monitoring, while generating publishable content remains widely discouraged. The Associated Press treats generative AI outputs as unvetted source material and does not permit ChatGPT to create publishable content. Jared Schroeder, associate professor at the University of Missouri, said students must be ready for newsroom policies that vary widely. "Whether their interviewer is very pro-AI or very skeptical of AI, our students can have an intelligent, thoughtful conversation with them about it," he said.

Lucas argued that AI literacy is as essential as fundamentals of communication law and democratic systems. "We have to know how government works, so we can keep it accountable. I think the same is true for the tech industry," she said. With Gallup reporting an all-time low in media trust last October, accuracy remains paramount, and unchecked generative AI output still falls short. Donna Davis, an instructor in public relations at the University of Oregon, compared the AI moment to the shift from analog to digital layout. "If we don't understand how to be the ones to harness that energy and use it responsibly, then we are all going to find ourselves in a much darker place than we're already in," she said.

This careful, discipline-level adaptation mirrors the broader challenge in education: preparing students for a changing workplace without diluting foundational knowledge. Resources such as AI for Education explore these integration strategies across subjects.

Classroom integration

Across all three schools, introductory writing and reporting courses remain AI-free. At Kansas, journalism professor Alyssa Appelman said her introductory writing class focuses on "introductory grammar, spelling, language, audience-centered writing skills" before anything else. Later courses introduce AI gradually. Lucas began teaching an AI elective in spring 2024, where students graded chatbots on journalism tasks, and in her data journalism course she shows students how to use AI to write code for analysis.

Appelman updated a fact-checking assignment to have students prompt AI to generate an article, then fact-check it. "Part of what's difficult about that is that any time we're introducing new skills into these classes, we're necessarily having to take some other skills out," she said. "Otherwise, we're just massively overloading semester classes."

At Oregon, professor Ed Madison's students spend a week in a "How Stories Work" course exploring AI's environmental, economic and philosophical angles. In another course, they learn to use AI - including Madison's own Sassy AI tool - to tailor resumes and cover letters for employers that increasingly rely on AI screening. Gary Kayye at the University of North Carolina has students produce two advertising campaigns for the same company, one with AI and one without, then lets the class vote without knowing which is which.

This hands-on integration requires instructors to become comfortable with AI tools themselves. For faculty developing those competencies, a program like the AI Learning Path for Teachers offers structured guidance. Madison is also pushing in the opposite direction: next fall he will lead a one-day workshop called "Adventures in Analog" where phones are surrendered and students practice observation, interviewing and sensory exercises with no technology at all.

Student perspectives

Not all assignments land easily. Schroeder had students prompt AI to write a paper on the First Amendment and felt he immediately lost them. Evaluations showed frustration and a desire for a human-driven education. "They said they hated it, but they learned something from it," he said. Some students simply refuse to use AI. "They're being told that the workforce is going to look a lot different, they're being told that it's going to be harder for them to gain employment … so why would they embrace that?" Madison said.

Appelman sees more reticence among undergraduates than graduate students, often out of fear of plagiarism accusations or messages from high school that AI equals cheating. One student refused to participate in an AI-related assignment because of the climate impact, so Appelman developed an alternative. Muzhingi said this skepticism isn't a problem to solve but a skill to nurture. "I actually see that skepticism as something which is healthy, because part of education is helping students critically evaluate these technologies, understand both the benefits and … those drawbacks, and in the end they kind of make their own well-informed decisions," he said.

Why this matters for higher education faculty

The journalism schools' path shows that AI integration works best when instructors have the autonomy to introduce it after core competencies are solidified. A top-down mandate is unnecessary; instead, faculty who know their students' readiness can time AI exposure to reinforce rather than replace foundational skills. The skepticism students bring - whether about ethics, accuracy or employment - can be channeled into critical-analysis assignments that turn ethical concerns into pedagogically rich moments, not obstacles. The goal is not AI fluency for its own sake but a graduate who can both report a story and interrogate an AI-generated source. This framework, which preserves essential human skills while layering in AI, offers a concrete model for curriculum design in any discipline confronting technological change.