What if everything we thought we knew about AI in education is wrong? Not the sci-fi stuff. Not the killer robots. I'm talking about how real students are using AI to learn, right now. We have data from over 10,000 student sessions with our AI tutor, and it points to a truth that nobody is talking about.
I've been in Silicon Valley for a long time. I've built and sold two companies, and I've written checks to over 200 startups, including some you've probably heard of like Anthropic, OpenAI, and Scale AI. I've seen my fair share of hype. And I've seen a lot of nonsense. I was getting tired of hearing “thought leaders” drone on about how AI was going to change education, all without a single shred of real data.
So, we decided to get some. We run an AI tutoring platform, which means we have a ton of data. We decided to dig into 10,000 individual student sessions. No theories. No buzzwords. Just the raw data of how students actually use AI.
The results? They surprised even me.
The Data Doesn't Lie: 3 Surprising Truths
We went in with our own biases. We figured we’d see kids trying to get quick answers for their homework. Maybe even trying to cheat the system. We were wrong. The data showed three clear patterns that go against almost everything the so-called experts are saying.
1. Productive Struggle is Everything
Everyone assumes an AI tutor should make learning easy. Just serve up the right answer on a silver platter. Our data shows the opposite. The students who learned the most weren't the ones who aced every question on the first try. They were the ones who struggled, failed, and tried again.
We saw a direct link between how many times a student got a question wrong and how well they retained the information a month later. It sounds backward, but it’s not. Real learning isn't about getting the answer. It's about the messy process of finding the answer.
When a student gets something wrong, our AI doesn't just spit out the correct answer. It nudges them. It asks probing questions. It forces them to think. That's where the magic happens.
It reminds me of my first startup. It was a spectacular failure. I made every mistake you can imagine. But I learned more from that one failure than from all my successes combined. The struggle is what forges you. It’s the same for these students.
2. Students Prefer the Robot
There's a huge push in EdTech to make AI tutors more... human. Give them a friendly avatar, a quirky personality, a sense of empathy. Our data says that's a terrible idea.
We found that students were far more willing to ask “stupid” questions and admit they were lost when they knew they were talking to a machine. There's no fear of looking dumb. No social pressure. They could be vulnerable, and that vulnerability was a superpower.
We even ran an A/B test. One version of our AI was the standard, slightly robotic tutor. The other was a super-friendly, human-like version. The results were not even close. Students using the “robot” were three times more likely to ask for help and spent 50% more time on the platform.
Think about it. When you’re learning something new, you don’t go to a world-class expert with your most basic questions. You go to a book, a video, a safe space where you can struggle in private. Our AI tutor is that safe space.
3. The “Forgotten Middle” Wins Big
The conventional wisdom says AI tutors are for the high-achievers who want to get ahead or the struggling students who are falling behind. Our data paints a very different picture. The students who benefited the most were the ones in the “forgotten middle.”
These are your B and C students. They’re not failing, but they’re not at the top of the class either. They’re the ones who get lost in a classroom of 30 kids. They don’t get the extra attention. They’re the ones most likely to slip through the cracks.
Our AI gives these students the one-on-one attention they’ve never had. It finds their specific weak spots and builds a learning plan just for them. And since it’s available 24/7, they can learn when and how they want.
I’ve seen this in business a hundred times. The most successful companies aren’t always the ones with the flashiest tech. They’re the ones that find an overlooked group of people and build something just for them. The “forgotten middle” is a massive group of underserved students, and AI is the tool that can finally unlock their potential.
Stop Listening to the Hype
So, what’s the takeaway? It’s time to ignore the gurus and look at the data. The real story of AI in education is not what you’ve been sold. It’s not about making learning easy. It’s not about making AI more human. And it’s not just for the top and bottom of the class.
It’s about embracing productive struggle. It’s about giving students a safe space to be vulnerable. And it’s about finally serving the underserved.
I’m more optimistic than ever about AI’s potential to change education. But we have to be smart. We have to follow the data, not the hype. The future of learning is here. It just doesn’t look like we thought it would.
Frequently Asked Questions
What experience informs this perspective?
This perspective comes from over a decade of building companies in Silicon Valley, two successful exits (RemoteTeam to Gusto, MovieLaLa to Gfycat), and investing in 200+ startups including Anthropic, OpenAI, and Scale AI. I write about what I've lived.
What's the most common pushback you get on this?
People often push back by citing exceptions or edge cases. And they're usually right that exceptions exist. But building a strategy around exceptions rather than patterns is a losing game for most founders.
How can I apply this thinking to my own situation?
Start by identifying the core principle behind the opinion, not the specific example. Then ask yourself: does this principle apply to my context? If yes, test it in a small, low-risk way before going all in.