Chipp Con 2025 Keynotes
In October we hosted over 100 people in Greenville, SC to share how they are using Chipp and AI to build the future of business and work. View all talks here.
Meet David
David Grewell is the Dean of the College of Engineering at Northern Illinois University (NIU). Before returning to academia, he spent 14 years at Emerson Electric, where he first started experimenting with AI and neural networks back in the late 80s and early 90s (when it took three years just to teach a machine to do anything).
At Chipp Con SC, David shared how he's tackling one of engineering education's biggest problems: the fact that 60% of engineering students drop out or switch majors. Watch his talk below 👇
The Real Problem: It's Calculus
Here's a stat that might surprise you: when a student enters any engineering college in the United States, only 40% will actually graduate with an engineering degree. The other 60%? They either switch to business or drop out entirely.
The main reason? Calculus.
David knew this when he joined NIU about two years ago, so he reached out to Scott with a specific goal: build an AI tutor that acts like a human tutor, not just another ChatGPT.
What Makes a Good Tutor?
David was clear from the start. He didn't want something that just spits out answers. He wanted something that guides students through problems the way a real human tutor would: asking questions, encouraging them, helping them think through each step.
Testing It Out
To assess the AI tutor, David got creative. He has a student success center with 10-15 human tutors on staff. When they weren't actively tutoring students, he asked them to test the AI tutor and their job was to play with it, try to break it, try to make it fail, and report back.
The results were pretty encouraging. The AI tutor got to the correct answer about 84% of the time, while human tutors hit around 80%. They're basically on par with each other, which was exactly what David was hoping for.
Students Actually Like It
David's team surveyed students about their experience, asking if they'd recommend it to friends. The response was overwhelmingly positive. Students found it helpful and would tell others about it.
Though David noted one interesting comment that kept popping up: "I'd rather use ChatGPT." And he gets it. ChatGPT is the easy way. You ask for the answer, it gives you the answer. But that's not learning. The AI tutor is designed to make you work for it, just like a good human tutor would.
Wrapping Up: AI That Teaches, Not Just Answers
David's approach shows what's possible when you think carefully about what problem you're actually trying to solve. He didn't want AI that makes things easier by giving answers. He wanted AI that makes learning better by guiding students through the process.
The fact that it performs on par with human tutors while being available 24/7 means students get support when they need it, not just during office hours. And if it helps even a portion of that 60% stay in engineering instead of dropping out, that's a win worth celebrating.
