Classes start at the University of Utah this coming Monday, August 24. Campus is getting busier and energy levels are rising. The start of a new school year is coming!
I was on sabbatical last academic year and although I did teach for a full semester at the University of Innsbruck, it feels like higher ed has changed dramatically since the last time I taught at the U in spring 2025. One reason for this is the rapid advancement of AI, including weather forecasting models and generative AI assistants like ChatGPT.
My approach to teaching has been one that has brought new tech into the classroom. The weather analysis and forecasting classes I teach today use tools and techniques that did not exist when I started as a professor and in some cases even a couple of years ago.
As we go into the new semester, I've been thinking a lot about how to utilize AI in the classroom but also how to ensure that students are using AI tools responsibly. The lowest-hanging fruit is bringing AI forecast modeling systems, like the ECMWF AIFS or Google WeatherNext into my lectures and weather discussions and using them alongside traditional numerical weather prediction model forecasts.
 |
| ECMWF AIFS forecast. Source: pivotalweather.com |
I'm not sure yet if I'll have time to develop my own graphics (my preference since I can tailor things for the western US and Utah) or add these systems to products like the Utah Snow Ensemble, but we will at least be using graphics from other sites.
I am more conflicted about generative AI and related tools that may have some potential to accelerate learning but can also be used in ways that prevent students from the hard work and deep thinking needed to develop the creativity and critical thinking skills.
Students need an advanced education that includes the foundational knowledge, expertise, judgement, and vision to extract maximum benefit from AI. There are no shortcuts to the top. A scientist needs a well trained mind just as an athlete needs a well trained body.
As an educator, I want to ensure that AI is a powerful assistant for learning and not a training shortcut that leads to cognitive atrophy. If I can borrow from a recent article by Mya Poe and Dylan Dryer, our students should be charting the path of scientific discovery, not machines.
Wonderfully said, Jim. Thanks for sharing. My wife will tell you this has been on my mind a LOT as I look at my final chapter of my dissertation. I'll be applying deep learning to some of our dynamical models, and there's definitely promise there. Specifically, the ability of CNN-LSTMs to see spatiotemporal patterns in these meteorological fields is so cool when applied to NWP.
ReplyDeleteObviously AI/ML is a huge blanket term and this kind of research is admittedly different than what's going on the industrial AI complex, but I've been conflicted about this for sure. As you mention here, I have very real concerns about how generative AI is impacting education and society as a whole, it's an enormous change we're all having to adjust to in a very short time. Personally, I've had to set hard boundaries on myself as a young scientist-in-training for how I interact with LLM's specifically, as it can be a huge temptation to "outsource" your thinking. Everyone's got to find what works best for them, but for me, I mostly just use the Google AI when I'm searching errors with coding and it's pretty quick at explaining it to me. I'm grateful that most of my education, I didn't need to worry about dealing with these hard questions, but I do have hope that incoming students will be thoughtful about how they interact with AI (maybe I'm naiive? haha). Anyways, just wanted to share my appreciation for your work in education and this site particularly, as a long-time fan/lurker on your blog and Utah snow in general. :D Crossing my fingers for a good ski season! Being a low to mid elevation XC skiier has really sucked these past few years...
Great to hear you are finishing up Casey. Thanks for the comments.
Delete