Dinowitz presses for empirical evidence; Dr. Chang cites studies

I mean, my general question still stands.

I just want to make sure you understand my question.

It seems like even if it were just this one school that we're using this generative AI text, our children are being guinea pigs.

And it just seems to me that I'm not hearing any studies or data that the DOE has done or that a third party has done or that peer-reviewed journals to demonstrate that these are tools that are beneficial to our students.

It seems to me that vendors have been approved and vendors are telling you that these are beneficial tools.

And I'm still not getting an answer to my question, which is what evidence is there that these tools are beneficial to our students?

So, Chair, I'd love to be able to respond to that question.

Please, thank you.

Because I'm...

In our review of the evidence, there are multiple meta-analysis, randomized control trials, in peer-reviewed scientific articles that state that

when these tools are being used with educators that are intelligent tutoring systems, that are structured systems, that are able to be in place,

What they're able to offer is really an ability to be able to provide careful supports around instructional design, feedback, and adaptability.

We want to be able to cite one such article that is in the journal PNJ, Science of Learning, that had 28 studies.

That was published in 2025.

There's another research article in PNAS that is a peer-reviewed randomized control trial.

Each of these pulled together many, many hundreds of studies and analyzed this research.

And this research ultimately tells us one thing, that it is really important that students are not having

unfiltered, uncontrolled, and unsupervised AI use.

But when it is being used together with teachers, under guidance and supported,

that challenges and asks students to be able to show their thinking and learning, that learning is helpful and does persist.