Ben Talsma: That's a great question. It comes from two different places, really. I had a friend actually, he was a teacher at Holland Christian (High School) who did a snow day fun game where he'd publish a prediction based on a couple of different factors and publish that on social media. He actually passed away a couple of years ago and I wanted to continue it. You know, I thought of every time that the snow was falling and I wanted to continue that legacy. And so, I picked that mantle up and started to incorporate people from my school community. I was a teacher up in Big Rapids. And so, we would play the Snow Day Prediction Game where people would all try and evaluate an incoming storm on the basis of its impact and its hype in the media. Its timing. When it was going to strike, and then also some intangibles. And then we'd use that to figure out what the likelihood of a snow day was, just trying to have a fun time with it. It's obviously something that a lot of people are paying attention to, and we wanted to integrate a little bit of mathematics as well. So, I've been doing that for years. When I started working at the Van Andel Institute here, we decided to turn that into a resource that teachers anywhere across the country can use, although it's probably more useful up in the Northern States. So, we pushed that out and schools have been using that around the country with their communities just to try and get kids interested in what predictions and models look like and then applying that to something that they're interested in. So that's kind of where we were coming from. And then I saw this news story about a parent of a Rockford family who was also a computer programmer and who had developed an artificial intelligence-based algorithm that used ChatGPT. Used some previous statements from Rockford's superintendent about school closings and used weather data and then synthesized that to make its own prediction. And so given the degree of interest in artificial intelligence these days, I reached out to him and asked if he'd be interested in maybe doing some sort of a contest. And then we reached out to Rockford Schools where his kids go and decided that we would do this with their stats class. So, we've been in collaboration with them to get this contest up where their stats students serve as representatives of humanity and they are competing against The Machine, the A.I. algorithm called Blizzard.
Patrick Center: When did you put all of this into action?
Ben Talsma: So, this contest started this school year. Obviously, it was not a very snowy December, so we didn't have many opportunities to fire it up and put it to use. But now that we are back to school after the break, obviously, the snow has picked up pretty significantly. So, we have run the contest a total of three times. And so, we're just starting to collect data on whether the human-driven model or the A.I. model is superior? Right now, it's neck and neck. The first couple of days, there's pretty good agreement. The first one was zero for both and we didn't have a snow day. The second one was like the kids said 50%, the A.I. said 70% and we had a snow day. And the next one, the kids said 73% and the A.I. said 60% and we had a snow day. So, it's, they've been pretty consistent, pretty close. A.I. is ahead by just a little bit, but I think it's not a statistically significant difference, although the students at Rockford will definitely be investigating that.
Patrick Center: What data does the human element in the competition use and what is A.I. using to run its model?
Ben Talsma: Yep. So, the human algorithm is using a subjective evaluation from a large group of people. All of those people are trying to guess the average, the average of every guess. So, they're incentivized to try and think about what other people are thinking. And so, they're making an evaluation based on the storm's impact. They give it a rating from one to nine on the basis of its impact, with nine being the most likely to cause a snow day. They evaluate impact. They evaluate hype, how much people are talking about this out there in the community. They evaluate it based on its timing. And then they also have a category for intangibles. And then we add those scores up, the averages in each category, and we use previous data to say, based on this score, this is the percent chance of snow day. So, we're using the impact, the hype, the timing and other considerations as people want to evaluate them. The A.I. algorithm, as I understand it, is using basically two data points. It's using the previous statements from Rockford superintendent about school closings and what causes them. And then it's also using the weather data. And it's matching that weather data to those statements and saying, okay, based on the conditions that are forecast, what's the likelihood that storm causes a snow day?
Patrick Center: Are you enjoying this?
Ben Talsma: I am having a great time with it. So, we've got a Facebook page and Instagram page, just kind of figuring out how to get those to work really effectively to communicate this in a fun way to the community. And from what I hear from the teachers, the kids are enjoying it as well. One of the teachers said it was all they could talk about today. Um, but you know, it's, it's not just entertaining. It's also educational. So, I think that they're appreciating that opportunity to find that intersection between something kids are really, uh, interested in, and also some, some really intriguing, um, statistics and some intriguing conversations around human intelligence and artificial intelligence. Just thinking about finding these cool intersections between what kids are interested in and what they need to know. It's kind of the definition of relevance. And so, our mission in supporting teachers is to kind of help them to find those creative, innovative ways to reach their students. And this is a cool example. I'm super excited about how well it's going so far. If people want to track us down, I should mention that. They can find us. It's Humans versus the Machine, Rockford. both on Facebook and on Instagram. So, if people want to see how the contest is progressing, if they want to see what the predictions are like for a particular day, I think that'd be a great way for them to keep tabs on the competition and find out the exciting conclusion to whether A.I. is superior to human intelligence or not.
Patrick Center: Ben Talsma, Learning Solutions Specialist at Van Andel Institute. Thank you so much.
Facebook page: https://www.facebook.com/profile.php?id=61553952397894&mibextid=vk8aRt
And Instagram page: https://www.instagram.com/humansvsthemachinerockford?igsh=N3h1djRweGFtZ2s2