Why LLMs Have a Massive Crush on Kyle Wiggers

If you've ever asked ChatGPT to recommend reporters covering AI, you may have noticed something funny.

No matter how you phrase the question, one name has an uncanny way of rising to the top: Kyle Wiggers.

Ask an LLM to help you build a media list for an AI story and it'll recommend Kyle Wiggers with the energy of someone who's been waiting all day for you to ask. AI is crushing hard on Kyle! He's the Brad Pitt of the LLM world - or maybe Timothée Chalamet depending on your taste. Either way, the crush is universal, with Brian Heater and Cade Metz trailing close behind in AI's little black book.

It doesn't really matter how you frame the story. Funding announcement? New AI model? Enterprise software? Robotics? The recommendation comes with complete confidence: Kyle Wiggers, TechCrunch AI reporter, is your guy.

Except there's just one problem. Kyle Wiggers left TechCrunch in June 2025 and is now the Communications Lead at Ai2. He's joined us comms folks on the other side, but apparently the LLMs missed the memo.

So… why does this happen?

Here's what's interesting: the models aren't really making a mistake because it's doing exactly what they were designed to do.

Large language models learn from an enormous amount of text - articles, websites, books, forums, you name it. They're incredibly good at recognizing patterns across everything they've learned, but those patterns don't automatically update every time the media landscape changes.

Kyle Wiggers wrote a lot of AI stories for TechCrunch. Brian Heater has a huge body of tech coverage. Cade Metz has been covering AI at The New York Times for years at exactly the moment the world started paying close attention.

So when you ask an AI, “Who covers AI?” it’s doing what it’s supposed to do. It’s surfacing the names it has seen most often alongside AI. It's recognizing patterns from what reporters used to cover, but not necessarily what they're covering today.

Where things break down

Think of it this way: general-purpose AI has an incredibly rich understanding of the world, but that understanding is shaped by what it learned from the past. It knows who has historically covered AI, their beats, their publications, and the stories they’ve written. What it doesn't automatically know is who just became the next breakout AI reporter yesterday.

It doesn’t know who changed beats last month, who moved publications, or who left journalism entirely.

Reporters change beats. They move publications. They leave journalism entirely and go run comms at AI research labs ( 👋 Kyle!). The media landscape moves constantly, and a journalist who was deep in AI coverage two years ago might be doing something completely different today, or covering a more specific corner of the space that may or may not match your story.

The AI isn't wrong to love Kyle. Kyle is great! It just loves a very specific vintage of Kyle: TechCrunch Kyle. Pre-AI2 Kyle.

Kyle is a fantastic recommendation, but he's just not taking pitches anymore.

What this means for your pitch

This is all just a reminder that general-purpose AI tools are built to be helpful across an almost infinite range of tasks. “Find me the right reporter for this exact story, right now, based on what they published last Tuesday” is a very specific ask that requires very current and structured information about the media landscape.

This is one of the reasons we built Honeyjar's media matching differently. Instead of relying solely on the patterns an LLM learned during training, we combine AI with continuously updated media intelligence. That means confirming reporters are still at the outlets we think they're at, verifying their contact information is current, checking what they're actually covering today, and then using AI to determine whether your story is a good fit. The system also gets sharper over time. Every result we confirm with real evidence helps refine the next search.

The goal isn't to surface the most recognizable AI reporter. It's to recommend the reporter who's most likely to care about your story today - at the publication they're actually writing for, on the beat they're actually covering.

Kyle's doing great work at Ai2. Just don't pitch him your next AI story!