Environment Canada says it is set to launch a first of its kind. Jean-Francois Caron, Research Scientist with Environment Canada, says they will soon be the first meteorological service on the planet with a public forecast that is partially generated using artificial intelligence (AI). Caron explains that weather forecasts are generated by humans. He notes Environment Canada will start using AI for issuing forecasts, but specifies that this will all be "upstream" from human intervention. Caron says forecasters base their forecasts on weather model outputs. He says historically, with these weather models, they are using the laws of physics to generate forecasts. "So, we provide (forecasters) the weather of the day, and with the laws of physics, they make forecasts up to several days," he adds. Caron says in recent years, AI models have started to change forecasting. He notes that instead of having the laws of physics coded in algorithms, forecasters are learning how the atmosphere works based on past weather. Caron says they are using decades worth of weather data to learn from. AI very good at long-range forecasting However, he points out that AI modeling is generally not as accurate as physics-based models. But where AI modeling has the edge is with large-scale weather phenomena. For example, Caron says for weather systems that cover areas of one thousand kilometres, the AI models are better than physics-based modeling. "However, they lack the fine-scale details, like high wind gusts, the high temperature, or very local precipitation," says Caron, referring to the inaccuracies of AI. "(AI) struggles to accurately predict those." So, Caron says Environment Canada will be implementing a hybrid model, which will use a mix of physics-based modeling and AI models. He notes that the large-scale weather phenomena will be guided by AI models, while the local forecasts with smaller-scale details will continue to use the physics-based models. "So, this is the big change that Environment Canada will implement in the month to come," he says. Humans will still be creating forecasts, not AI But Caron emphasizes that the role AI will play in forecasting is different from how the average person uses AI. "When you say AI, people think about ChatGPT," he says. "Here, it's something completely different." Caron explains that AI will not be generating text for Environment Canada. Rather, it will generate data, which is then interpreted by human beings in order to make the public forecast. Caron says they have spent more than one year doing research and development in order to understand the pros and cons of this new approach. He notes this is how they have come to realize that AI lacks in the small-scale forecasting but is very good with large-scale forecasting. "This is when we had this idea to do a blend of the two approaches," he says. "So, we will be the first weather service in the world to implement this." Official launch expected this spring Caron says Environment Canada is anticipating the launch of its AI-influenced forecasts during the second half of May. He notes that in terms of the science and the quality of the forecast, they are happy with how things are working and have the green light to proceed. Caron says it is just a matter of making sure that everything is ready from a technical perspective. According to Caron, Environment Canada's forecast, updated several times daily, will not look any different after they launch this change. The change is in the method used in putting together the forecast. And the part of the forecast that will be created with help from AI is for days three to seven. "It's really from day three to day seven for the public; that there will be big gains in terms of accuracy," adds Caron. "We're saying that a six-day forecast with the new model will be as good as a five-day forecast from the previous model for us; it's a huge change, and we are very proud." Environment Canada expecting improved accuracy for major systems Environment Canada says this change will improve their large-scale accuracy. This is expected to translate into predictions of major systems such as winter storms, heat waves, or atmospheric rivers from eight hours to more than 24 hours earlier. Environment Canada says there will also be improved accuracy for predicting when specific weather conditions are expected to start and for mapping the track that a storm will follow. Meanwhile, for those concerned that the use of AI will replace the need for meteorologists, Caron says there is no need to worry. He notes this change will not take jobs away from meteorologists. In fact, he says it will actually create jobs on the research side of things.