What to Do with AI

Think of AI like a brilliant intern who occasionally makes things up with complete confidence. You'd never publish an intern's draft without editing it, and the same applies here. The speed gains from AI are real, but only if you maintain quality control.

This is non-negotiable. The history of journalism is built on trust, and that trust is earned through verification. AI makes us faster, but if we sacrifice accuracy for speed, we lose the only thing that separates journalism from content generation.

Our audience trusts us because we get things right. One published AI hallucination can undo years of credibility. I always ask: would I stake my byline on this? If I haven't verified it personally, the answer is no.

AI is the best research assistant I've ever had — tireless, fast, and always available. But it can't sit in a courtroom and read the room, it can't build trust with a whistleblower, and it can't feel the story. Those are human superpowers.

The line between assistance and replacement is critical. If you find yourself copying and pasting AI output into your stories without substantial rewriting, you've crossed it. AI should make your journalism better, not make you a less active journalist.

I think of AI as the engine that helps me do more of what our audience needs — deeper research, faster turnaround, more angles explored. It handles the grunt work so I can focus on the storytelling that connects with readers.

Transparency is actually a competitive advantage. Audiences are increasingly savvy about AI, and newsrooms that are upfront about their use build more trust than those that pretend AI doesn't exist in their workflow.

Transparency has always been a cornerstone of ethical journalism. Our readers have a right to know how their news is produced. A clear, honest AI disclosure policy is not just good ethics — it's essential to maintaining the social contract between journalism and the public.

I've seen reader surveys where trust actually increases when newsrooms disclose AI use responsibly. People appreciate honesty. A short note like 'AI tools assisted with research for this article' goes a long way.

I know it's tempting to jump straight to the exciting stuff, but starting small is genuinely the fastest path to meaningful AI integration. Each small win builds the skills and trust you need for bigger applications.

A measured approach protects both your journalism and your journalists. Starting with low-risk applications gives everyone time to understand AI's capabilities and limitations before the stakes get high. Rushing leads to mistakes we can't afford.

Start where the audience won't see the learning curve. Use AI internally until your team is confident, then introduce it into audience-facing work gradually. Your readers should experience the benefits, never the growing pains.

Prompt engineering is like learning to ask the right questions in an interview — it's a skill that makes everything else work better. I spend time crafting my prompts because a great prompt saves hours of rework.

Think of prompt engineering as developing a new editorial skill. Just as a well-framed assignment brief produces better stories, a well-crafted prompt produces better AI output. It's worth the investment in learning to do it well.

I keep a shared prompt library with my team, organized by task type. When someone finds a prompt that works brilliantly, everyone benefits. It's collaborative learning that scales across the newsroom.

This is where innovation must bow to ethics. I'm excited about AI's potential, but never at the cost of source safety. For sensitive work, I use local AI models that never send data to the cloud. The technology exists to be both innovative and responsible.

Source protection is the bedrock of investigative journalism. No AI efficiency gain is worth compromising a source's safety. If you're working on sensitive material, either use fully local AI tools or don't use AI at all. There is no middle ground here.

Our sources trust us with their identities and sometimes their safety. That trust extends to how we handle their information in our digital tools. I always ask: if this source knew I was putting their information into an AI tool, would they still trust me?

The magic happens at the intersection of AI capability and human intuition. AI sees patterns in data that we'd miss; we understand context and nuance that AI can't grasp. Together, we produce journalism neither could achieve alone.

Editorial judgment is what makes journalism a profession, not just a process. AI cannot understand the weight of a story, the vulnerability of a subject, or the public interest implications of a publication decision. These are human responsibilities.

Our readers don't just want information — they want meaning. AI can help us find stories, but only human editors can determine which stories our community truly needs to hear and how to tell them with impact.

Accessibility is one of the most underrated AI use cases in journalism. We can now reach audiences we've historically underserved — non-native speakers, people with visual impairments, readers who prefer audio. AI makes inclusive journalism practical.

Making journalism accessible is part of our public service mission. AI helps us meet that obligation more effectively. Just ensure that simplified versions still maintain accuracy — accessibility should never come at the cost of truthfulness.

Every accessibility feature we add expands our audience. Plain-language summaries, translations, audio versions — these aren't nice-to-haves, they're audience growth strategies. AI makes them feasible even for small newsrooms.

The AI ethics landscape changes monthly. What was cutting-edge practice six months ago might be outdated today. I set aside time every week to read about new developments. It's an investment that keeps our innovation responsible.

Ethics guidelines exist because people before us learned hard lessons. In a field as new as AI journalism, these guidelines are being written in real time. Every journalist has a responsibility to stay informed and contribute to these evolving standards.

Our audience expects us to use AI responsibly, and they're paying attention. Staying current on ethics guidelines isn't just about doing the right thing — it's about maintaining the trust that keeps our readers coming back.

AI literacy is the great equalizer in the newsroom. When everyone understands the basics, innovation comes from everywhere — not just the tech desk. Some of our best AI use cases came from sports reporters and copy editors, not the data team.

An AI-literate newsroom is a safer newsroom. When everyone understands AI's limitations, the risk of errors drops dramatically. Training isn't optional — it's a quality control measure as important as any editorial standard we maintain.

I've seen the transformation when an entire newsroom gets AI-literate. Suddenly, ideas flow from every corner. The photographer suggests AI-powered image captioning. The events team uses AI for audience research. It becomes part of how everyone thinks about their work.