How to Believe in AI and Education Without Really Trying
Believing in AI in education is easy. Building something that actually works is hard. The gap between the two is where most ed-tech goes to die.
It is very easy to believe in AI in education. You watch a demo. The founder is passionate. The slides are beautiful. The pilot results are promising. You leave the meeting thinking, "This could change everything."
And then nothing changes.
This pattern has repeated itself in ed-tech for twenty years. Interactive whiteboards were going to change everything. MOOCs were going to change everything. Learning management systems were going to change everything. Adaptive learning 1.0 was going to change everything.
The technology was often genuinely impressive. The demos were often genuinely compelling. And the implementations were almost always disappointing. Not because the technology failed, but because the integration failed.
Here is what I mean by integration failure: the tool works in the demo but does not fit into the faculty member's actual workflow. The data is interesting but does not connect to the decisions the instructor actually makes. The students use it for a week and then stop because it feels like extra work rather than a natural part of learning.
Belief is not the hard part. Integration is the hard part.
When we built Arrival, we started with the workflow, not the technology. We asked: what does a faculty member actually do on a Tuesday afternoon when they are worried about a student? They look at grades. They check attendance. They maybe send an email. They wait for the midterm and hope for the best.
Arrival does not ask them to change that workflow. It gives them better information within it. Instead of looking at a grade that says C-minus, they see a proficiency map that says: this student understands cellular mechanisms well but cannot connect them to tissue-level function, the trajectory has been flat for three weeks, and here are two specific concepts where targeted practice would move the needle.
That is not a revolution. It is a better Tuesday afternoon. And better Tuesday afternoons, compounded over a semester, are how education actually changes.
Stop believing in AI in education. Start demanding that it work on a Tuesday afternoon.