Jeremy Siegel: This is GBH's Morning Edition. There's a prevailing message from experts and leaders in technology that artificial intelligence will change everything: Medicine, film and TV, journalism. We've heard a lot about how AI could upend these industries. But one field we haven't talked too much about is the weather. Not that AI will change it, but how it can predict it. Newer AI models can forecast the weather up to 10 days in advance in less than a minute. And joining us now to talk more about this and what it means is our meteorologist Dave Epstein. Good morning Dave.

Dave Epstein: Good morning Jeremy. Good to be here.

Siegel: It's good to have you. So I just got to ask off the bat, have you used any AI software with weather?

Epstein: No, I've peeked at it. So, you know, everybody has heard pretty much about the Euro and the GFS and all those different models. And there is an AI version out there that you can look at on some of the subscription services that we get. So you can peek at, you know, what does the AI model say about a particular storm? And I've looked at it and in some cases it is performing, you know, okay. And we'll talk about, you know, some of the caveats about that. But yeah, I've glanced at it, but it's not in widespread use yet.

Siegel: Well, let's talk about how well it works. I mean, how accurate is it when you've seen AI predictions, and what are the caveats?

Epstein: Yeah. So here's the big difference. So our regular models, the Euro, the GFS, they basically take data that is assimilated from weather balloons every single day, twice a day. They take that real time data and they say, okay, here's the state of the atmosphere right now. What's it going to look like in an hour, in 24 hours and 48 hours, all the way up to 10 days in some cases, and even longer. And so they're taking that real data and looking at it. And it's a very computer-intensive process. It takes hours, so 2 to 3 hours from the time you start ingesting the data until you get the forecast. With the AI, it's saying, okay, here's what the atmosphere looks like. In the past, what's happened? Right? So it's saying, basically not looking at the real data in the same way, it's just saying, okay, a storm off the coast of the Carolinas with high pressure here and low pressure there and the upper levels looking like — this did this in the past. And that's what it predicts. And so it's based on past weather, not what's really going on now. Now obviously it has some of the real weather because the past weather has real weather. But it's very fast. You know, as you said at the beginning, it's just a matter of minutes to get a forecast, in that case.

Siegel: How different is what an AI model is doing to what you're doing every day when you're here on GBH's Morning Edition predicting the weather?

Epstein: Yeah. I mean, the AI model, like if I showed you an AI prediction and showed you the Euro prediction, you would say, I don't really see the difference. The position of the storm might be different, but it's not really that different. It's how the prediction's made with the AI. So GraphCast, which was developed by Google DeepMind, is one of latest several AI weather models. And you know, they're looking at, okay, how does that AI model predict what's going to happen with this particular storm? But again, it's not based on the atmosphere on March 28th. It's based on all the historical stuff that's occurred, and based on that, here's what probably is going to happen. It's a much narrower forecast. Like when I look at the euro, I get: There's a probability of this and a probability of that. And I can say, okay, what's the highest probability? What's the mid probability? What's the lowest probability? You can't do that yet with any of the AI stuff.

Siegel: Looking down the road though, do you foresee a future where you yourself as a meteorologist who right now is using these different models to make your predictions about the weather, do you think in the future, someday you're going to be using artificial intelligence to inform your forecasts?

Epstein: I definitely think that it's going to be part of the suite of data that we all use to think about how forecasts are going to be made, and I think that over time, we will gain more knowledge as to just how accurate the AI models are. Just like today, I'm going to rely more heavily on the Euro, because it's a more accurate model, over the American GFS. So, you know, you fast forward three, five, seven years, what have the AI models been doing? How well have they been performing compared with our more conventional operational and ensemble models?

Siegel: Is there a concern that comparison will also be how well is AI doing next to a meteorologist?

Epstein: Well, I think you're still going to have to interpret it. And the thing about the AI is that it's using past data, right? And so as we go into more extreme weather and we see things like 60 inches down in Houston during that hurricane a few years ago, the AI doesn't have, since it's using past stuff, there's nothing that's occurred like that in the past, so it's going to have trouble predicting those more extreme events because it doesn't have an extreme event in it's sort of historical knowledge, right? So that's where the operational models and an actual human being are really going to come into play.

Siegel: That is GBH meteorologist and actual human being Dave Epstein. Dave, thanks so much for your time this morning.

Epstein: You're welcome.

Siegel: You're listening to GBH News.

As the tech industry seeks more uses for artificial intelligence, some people have turned their attention to weather forecasting.

Newer AI models like Google’s GraphCast can generate a 10-day weather forecast in minutes, GBH meteorologist Dave Epstein said.

But while those forecasts can be accurate, they also have limitations — especially in a changing climate.

To understand how they work, Epstein said, it’s important to first take a look at the models he and other meteorologists use now.

“Everybody has heard pretty much about the Euro and the GFS and all those different models,” he said. “If I showed you an AI prediction and showed you the Euro prediction, you would say, I don't really see the difference. … But again, it's not based on the atmosphere on March 28. It's based on all the historical stuff that's occurred, and based on that, here's what probably is going to happen. It's a much narrower forecast.”

The current models feed real-time data collected from weather balloons into algorithms to try and predict what conditions might look like in the near future — the next hour, the next day, the next 10 days, he said. It’s a process that can take a while.

“It's a very computer-intensive process. It takes hours, so two to three hours from the time you start ingesting the data until you get the forecast,” Epstein said.

AI models, in contrast, compare current conditions to historical data and produce an output of what’s happened in the past when conditions were similar, he said.

“It’s basically not looking at the real data in the same way,” Epstein said. “It's just saying, okay, a storm off the coast of the Carolinas with high pressure here and low pressure there and the upper levels looking like this — did this in the past. And that's what it predicts.”

The upside: A forecast that takes minutes instead of hours to generate. The downside is that the AI models can only use previously-recorded data that has been fed to them, so they may not be able to forecast unprecedented weather events, like Hurricane Harvey, which devastated parts of Texas in 2017.

“As we go into more extreme weather and we see things like 60 inches down in Houston during that hurricane a few years ago, … there's nothing that's occurred like that in the past, so it's going to have trouble predicting those more extreme events because it doesn't have an extreme event in it's sort of historical knowledge,” he said. “That's where the operational models and an actual human being are really going to come into play.”

Epstein said he looks at forecasts not just for the most likely scenarios, but for mid-range and low-range probability conditions.

“You can't do that yet with any of the AI stuff,” at least at this point, he said.

Though the AI models are not yet in widespread use, he said he is keeping an eye on new developments.

“I definitely think that it's going to be part of the suite of data that we all use to think about how forecasts are going to be made,” Epstein said. “And I think that over time, we will gain more knowledge as to just how accurate the AI models are.”

And he doesn’t expect meteorologists will be out of a job.

“I think you're still going to have to interpret it,” he said.