Jean-Marc Apollon of NYC Bar Association testifies on AI-assisted grading pilot proposal
Jean-Marc Apollon of the NYC Bar Association argues a complete ban on AI for grading may create 'shadow AI' problems, and recommends a limited pilot program where AI recommends and teachers decide, with careful measurement of results and bias.
Good afternoon, chairs and members of the committee.
My name is Jean-Marc Apollon.
I'm the co-chair of the New York City Bar Association's Technology, Cyber, and Privacy Law Committee.
Thank you for the opportunity to testify today.
We appreciate that New York City Public Schools is taking a careful approach to artificial intelligence.
AI present real concerns, including bias, inaccurate information, and student privacy.
But we believe the current complete ban recommended by the guidance on artificial intelligence and screen time on teachers using AI for grading goes too far.
On one hand, a complete ban can create a different problem, shadow AI.
AI is becoming part of everyday life and work.
If teachers find these tools useful but are told they cannot use them, some may turn to their own personal and unapproved AI systems.
That could mean student work or personal information being entered into systems that the city has not reviewed and cannot oversee.
On the other hand, we do not want the opposite extreme either.
Teachers should not remove themselves from the process and let AI grade students or run the classroom.
Teachers know their students.
They understand context that an AI system may not.
They should remain responsible for every final grading decision.
NIDR Extreme is the solution.
The goal should not be to introduce AI simply because the technology exists, and the goal should not be to prohibit AI simply because the technology presents risks.
The goal should be better education.
That is why we recommend a limited pilot program for AI-assisted grading.
The basic rules should be simple.
AI recommends, teachers decide.
Start small, select a limited number of schools and teachers, use only AI tools that have been reviewed and approved by the city, then measure the results.
Did teachers save time?
Did students receive better feedback?
Were grades more consistent?
Did the technology treat different groups of students differently?
How often did teachers disagree with AI?
And most importantly, did it improve the educational process?
New York City has an opportunity to show that protecting students and embracing innovation are not opposing goals.
We can protect student privacy.
We can address bias.
We can keep teachers in control.
And we can still explore whether new technology can improve education.
We should not make decisions based on fear of AI or excitement about AI.
We should test it, measure it, and build the policy based on what the evidence tells us.
Thank you for the opportunity.
also reflects both of the co-chairs' opinion of having evidence-based policies and not just emotional or reactive policies.
Thank you so much.