The research that validates what you already suspected about AI in classrooms

If your gut has been telling you that most AI tools aren't actually good for student learning, the research is starting to back you up.

Here's what we're seeing: when students use AI to complete their work, they often produce a better looking finished product, but the same regions of the brain aren't lighting up as they would if the student had done it themselves. And, there's no skill transfer. Doing something better with AI ten times doesn't mean they’ll be able to do it on their own the eleventh time.

There's a name for this kind of AI, and I think it's the perfect description: sycophantic. It tells kids their ideas are brilliant. It validates wrong answers. It does the cognitive heavy lifting so the student doesn't have to. And, it's the dominant model right now in consumer AI tools, because engagement and flattery keep people coming back, which is great for a tech company's metrics and terrible for a kid's brain development.

The research that's showing more promise looks completely different. It's what you'd recognize as Socratic. The AI pushes back and asks the student to explain their thinking. It asks questions rather than giving the answers, creating productive struggle rather than eliminating it. Socratic types of AI will interact with students in ways that mirror effective teaching techniques that you use everyday.

The problem is that most of the AI tools being pitched to your district right now are the sycophantic kind, and the decisions about which tools end up in your school are often being made without teacher input. Additionally, these tools are being adopted without clear evidence that they actually improve student outcomes, often by disregarding what we know about the science of learning.

I went deep on all of this in my latest podcast episode. I interviewed Erin Mote, CEO of InnovateEDU,a nonprofit that's been at the forefront of responsible AI in education for years. She convened the SAFE AI Companions Task Force, which brought together over 70 educators, researchers, policymakers, and tech developers (including me, on behalf of 40 Hour AI) to set guidelines for how AI should and shouldn't be used with kids.

You can read or listen to the conversation here:

AI in the classroom: what the learning science actually says

And if you want to go even deeper, I'm running a two-day virtual training on July 13 and 14 called Stay Human: Finding Your Ethical Stance on AI as an Educator.

The purpose is to help you take a step back from the in-the-moment decisions you have to make during the school year, and get clear on where you stand, what your values are, and how to make consistent decisions about AI in your teaching.

We'll cover your own AI use as a teacher and your students' AI use, with the learning science woven throughout.

You'll walk away with a personal and classroom AI policy that's aligned to both your district's mandates and your own teaching philosophy, plus two ready-to-use mini units to help your students think critically about AI.

Live sessions run 10 am to 12:30 pm Eastern time on both days, so you don’t have to get up super early, are done by lunch, and have time in between to let the ideas marinate.

All the recordings are included, and you can earn up to 7 hours of PD credit.

Regular price is $97, but you can grab the early bird price of $47 now. (If you're already a member of 40 Hour AI, this is included for you at no charge.)

Register for Stay Human July 13-14

Enjoy!

Angela