Case Study

Lenfest AI Fellow spotlight: Kevin Hoffman

The Philadelphia Inquirer

October 19, 2026

This profile is part of the Lenfest AI Collaborative and Fellowship Program’s 2026 Impact Report.

As an AI Fellow at The Philadelphia Inquirer, Kevin Hoffman is helping a 197‑year‑old newsroom navigate a fast‑moving AI landscape while staying grounded in editorial values and real‑world workflows. In conversation, he reflected on how his role has evolved and how another Lenfest AI Fellow helped him build trust with reporters.

About his role and the problem he is trying to solve

Kevin joined The Inquirer at the end of 2024, initially with one clear goal: To build a digital librarian. He worked closely with the Microsoft AI Development Acceleration Program (MAIDAP) team, including a two-week hackathon with their engineers.

“We quickly realized that we were able to prototype a lot of these more complex applications far quicker than anticipated, just through the resources made available to us in the grant through the MAIDAP team at Microsoft,” he said.

In the early months, the fellowship was highly engineering-focused. Kevin spent his time with his head down, building prototypes and showing tools to stakeholders to get them to buy in.

“The role has largely transformed throughout the year and a half,” he said. “I have definitely taken on a more considerable role with also being like an AI liaison or consultant with the newsroom.”

That bridge role includes working with leadership on internal AI policies, talking with reporters about their experiences with the tools, and increasingly, helping reporters build their own tools as they become more comfortable with AI systems.

“That’s been the big topic recently,” Kevin said. “I hear from lots of different staff in the newsroom, like, ‘Hey, I vibe coded this project, I don’t really know much about coding, but it really helps me, and I’d like to share it with my coworkers. Is there any way we could build this into production and have multiple people collaborate on this project?’”

His flagship project

Kevin’s original digital librarian project, Dewey, is now moving into its next phase. He is focused on two expansions:

Image search over the Inquirer’s photography archives, so reporters can search as easily for visuals as text.
Exploring digitizing historical newspaper issues that pre-date the existing digital archives, with the goal of introducing much older material into a searchable system.

He is also supporting another tool that The Inquirer developed, a local media aggregator called Scrape, designed for journalists working on hyper-local newsletters for townships around Philadelphia.

Reporters would previously spend 12 hours a week checking hundreds of sources, such as government and school district sites, for potential stories. “It was a real bottleneck,” Kevin said.

Now, reporters give Scrape a list of sources to monitor and receive a daily report summarizing updates from the last 24 hours.

“We’ve really tried to simplify the news aggregation process from a 12-hour task a week to something that is more like opening your inbox and gathering tips from Scrape’s report,” Kevin said.

The biggest surprise during the fellowship

Kevin went into the role expecting trust-building to be a major challenge. What surprised him most is how quickly newsroom needs drove ideas for AI tools, once the usefulness of AI — and trust — had been established.

“Once we had crossed that bridge, just how fast the pace would move and how much conversations would shift surprised me,” he said. Adoption went from journalists asking whether they should use AI to complaining that a tool was down for even 30 minutes. “I didn’t realize it was being depended on like that.”

The real accelerator came when reporters became vocal advocates for AI tools.

“We’d gotten really close with several different investigative reporters that had strongly evangelized the use of some of these tools within the newsroom,” he said. “And building that level of trust with people with deep institutional knowledge at The Inquirer really accelerated that pace.”

One thing the industry gets wrong about AI

For Kevin, one widespread mistake is the way AI has been rolled out.

“The way that AI has been deployed across industries, not just the news industry, has been kind of in everybody’s face,” he said. “It’s in Zoom, it’s in Slack, it’s in Gmail, it’s kind of everywhere you go, and that can feel really exhausting.”

AI was often sold as a tool that can do anything, leading to requests that don’t match what large language models are actually good at.

“We were finding that a lot of people would ask questions like, you know, ‘Give me every single time we’ve ever mentioned this one person in our reporting, and I would like you to figure out how many articles we’ve written about them that’s positive,’” he said. “That’s a really hard thing for AI to do… it’s not a tool that’s meant to be used as an exhaustive feature.”

Instead, he emphasizes the need for AI literacy and clear communication about the limitations and appropriate uses of these tools.

“Once we are able to arrange them in a way where people understand exactly where the sort of pitfalls are, we get a lot further,” he said.

Lessons other newsrooms could apply tomorrow

One of Kevin’s most practical lessons is about showing up and listening.

He credits ProPublica AI fellow Dana Chiueh, a reporter-turned-engineer, with pushing him to spend more time simply talking with journalists about how they felt about AI before pitching solutions.

“Have you made an effort to go out there and connect with the journalists?” he recalls her asking.

That advice helped him see that building effective AI tools for journalism meant immersing himself in editorial workflows. He believes publishers should embed people at the intersection of engineering and the newsroom, and then give them the time and support to deeply understand pain points, constraints, and values.

“Building AI as an AI engineer doesn’t mean much unless you have a subject matter expert in the area you’re trying to build AI for,” he said.

As a loyal Inquirer reader but a news industry newbie, he needed to learn “what it looks like for high-quality journalism to be produced” and where tools could genuinely help.

What he is paying attention to right now

Explainable AI, a field in which Kevin worked extensively before this fellowship.

“We’re increasingly becoming reliant on a lot of these tools that are AI-focused in so many areas, but we still feel like they’re a black box,” he said.

What he is most optimistic about

Kevin marvels at how he is witnessing the cultural evolution underway in the newsroom, with reporters growing more comfortable with AI, becoming builders themselves, and using AI tools to support serious investigative work.