Every newsroom is doing AI training now. Workshops, webinars, lunch-and-learns, mandatory online modules.

Most of it doesn’t work.

I’ve observed AI training programs at more than two dozen news organizations over the past two years. The gap between effective programs and useless ones is enormous. Here’s what I’ve learned about the difference.

Why Most Training Fails

The typical AI training session goes something like this:

A presenter demonstrates ChatGPT or Claude. They show how to write a prompt. They demonstrate a few use cases. Attendees nod. The session ends.

Two weeks later, almost nobody has changed how they work.

The failure modes are consistent:

Too abstract. General AI concepts don’t translate to daily work. Journalists need specific applications for their specific tasks.

Wrong audience. Mixing technical staff with complete novices frustrates everyone. Differentiated training matters.

No practice time. Watching demonstrations doesn’t build skills. People learn by doing.

No follow-up. A one-time session can’t create lasting behavior change. Adoption requires sustained support.

No workflow integration. Learning about AI in a vacuum doesn’t help. Training must connect to actual newsroom processes.

What Effective Training Looks Like

The successful programs I’ve observed share common elements:

Role-Specific Applications

The best training starts with the journalist’s actual job and works backward to AI tools.

For a beat reporter: How can AI help you monitor your beat? Research background? Identify sources?

For an editor: How can AI assist with headline testing? Content planning? Quality control?

For a data journalist: How can AI help with data cleaning? Pattern identification? Visualization?

Generic training produces generic adoption. Specific training produces actual use.

Hands-On Practice

Every effective program includes substantial practice time—not demonstrations, but participants actually using tools on their own work.

One program I observed dedicated 70% of session time to hands-on work. Participants brought real assignments and worked on them with AI assistance during training.

Another required participants to complete specific tasks between sessions—not optional homework but tracked assignments.

The programs that work understand that skill-building requires repetition, not just exposure.

Peer Learning

Some of the best training happens between colleagues, not from formal programs.

Smart newsrooms identify internal AI champions—journalists already using tools effectively—and create structures for them to share knowledge.

This peer approach has advantages:

• Champions understand the specific newsroom context
• They’re available for ongoing questions
• They have credibility with colleagues
• They learn by teaching

The formal training matters, but the informal peer network sustains adoption.

Ongoing Support

Behavior change doesn’t happen in a single session. Effective programs include:

• Follow-up sessions addressing questions that emerge with use
• Office hours or drop-in support for troubleshooting
• Slack channels or other forums for peer discussion
• Regular refreshers as tools and capabilities evolve

The investment in ongoing support often exceeds the initial training investment. That’s appropriate.

Clear Policies

Journalists won’t use tools if they’re unsure what’s permitted. Effective training includes:

• Clear policies on acceptable use
• Examples of appropriate and inappropriate applications
• Guidance on disclosure and labeling
• Escalation paths for ethical questions

Ambiguity about rules creates hesitancy. Clear guidelines enable confident adoption.

Common Mistakes

Beyond the generic failure modes, I see specific mistakes repeatedly:

Starting with technical explanations. Journalists don’t need to understand how large language models work. They need to know what the tools can do for them. Lead with applications, not architecture.

Overselling capabilities. Training that makes AI sound magical sets up disappointment when reality falls short. Be honest about limitations.

Ignoring skeptics. Some journalists are genuinely worried about AI’s implications for their profession. Dismissing these concerns creates resistance. Acknowledge them honestly.

Treating all tools equally. ChatGPT, Claude, Gemini, specialized journalism tools—they differ in meaningful ways. Training should help journalists understand which tools work best for which tasks.

Skipping the “why.” Some journalists will resist AI simply because change is uncomfortable. Training needs to make the case for why AI adoption matters—for their productivity, their journalism, their sustainability.

Building a Training Program

If you’re developing AI training for a newsroom, here’s a framework:

Assess current state. Survey staff on current AI use, comfort levels, and specific needs. Different starting points require different approaches.

Differentiate by role. Create separate tracks for different functions—reporting, editing, production, data work. Generic training serves nobody well.

Start with quick wins. Begin with applications that are immediately useful and low-risk. Transcription. Research assistance. Routine drafts. Build confidence before tackling complex applications.

Provide resources. Written guides, prompt libraries, policy documents. People need reference materials, not just sessions.

Measure adoption. Track whether training translates to actual use. If adoption isn’t happening, the training isn’t working.

Iterate. Training programs should evolve based on feedback and results. What’s working? What’s confusing? What’s missing?

External Help

Some newsrooms have internal capacity to develop and deliver training. Many don’t.

External resources include:

• Industry associations offering AI training programs
• Journalism schools with continuing education offerings
• Consultancies specializing in newsroom AI, like team400.ai
• Peer newsrooms willing to share materials and approaches

The build-versus-buy decision depends on scale, budget, and internal expertise. For many newsrooms, some external support accelerates the process considerably.

The Skeptics

A note on skeptical staff: they’re not wrong to be cautious.

AI does pose real questions about journalism’s future. Concerns about job displacement, quality erosion, and ethical risks are legitimate. Training that pretends these concerns don’t exist will fail.

Effective training acknowledges the concerns honestly while making the case for engagement:

• AI is coming regardless. Better to shape its use than ignore it.
• Skills in using AI are increasingly valuable professionally.
• Thoughtful AI use can improve journalism, not just reduce costs.
• Engagement gives journalists a voice in how these tools are deployed.

Some skeptics will remain skeptical. That’s fine. The goal is informed skepticism, not forced enthusiasm.

Measuring Success

How do you know if training is working?

Adoption metrics. Are journalists actually using AI tools? Track usage if possible.

Quality indicators. Is AI use improving work product? Faster production? Better research?

Attitude surveys. Has comfort with AI increased? Have concerns decreased?

Anecdotal feedback. What are people saying about their AI use? Collect stories.

Workflow integration. Has AI become part of standard processes, not just occasional experimentation?

Training that doesn’t produce measurable adoption isn’t working. Be willing to change approaches based on results.

Looking Forward

AI capabilities will continue expanding. Training programs must evolve accordingly.

The newsrooms that treat AI training as an ongoing investment—not a one-time project—will adapt more successfully.

Building internal capacity matters. Working with a Sydney-based firm can help develop sustainable training programs that evolve with the technology.

AI training isn’t just about tools. It’s about building organizational capacity to adapt as technology changes. The newsrooms that invest in this capacity now will be better positioned for whatever comes next.

---

I’m compiling examples of effective newsroom AI training programs. If your organization has developed an approach that’s working, I’d love to learn about it.