Mark: No surprise here. We are still in the early stages. But we already see some very promising developments.
Someone said something that's stuck with me recently: “AI is as bad as ever. It's just getting better.” I think it's a great way to see it. In the overture, we consider the role of AI in two ways.
First, how to help end users access and use the map. That's what most people think about. But secondly, and more importantly for us, is how AI can help us build, validate and improve our mapping data. That's the backend side where we invest a lot. Historically, map data came from research.
Back in Lewis and Clark, or Spanish and European explorers, they were out on the field, observed and recorded. And until just a few decades ago, the best maps were recently researched.
Then the phone recognized the location and suddenly the sensors (billions of sensors) were giving us information. This has led to the concept of remote sensing. Observe and infer data using satellites, phones, or other devices. It has been a major evolution over the past 15 years.
This enters a new phase where large-scale language models and AI agents can help create or validate data.
Let me give you an example: If you ask the example “Is Keith's burger restaurant in Elko, Nevada still in 2025?”, the model might search for reviews, social media, or other metrics and say, “Yes, I saw an activity that suggests it's still open.” It's powerful.
In the past, someone has to physically send it to you to confirm it. Now, agents can repeatedly check various sources to streamline and verify the data by bringing together signals. I think it will revolutionize the way maps are built.
