A Lesson is Repeated Until It Is Learned

AI

Written by: Christian Talbot, MSA President, MSA (posted Friday, August 1st, 2025 | 4:53 pm)

Photo by Shubham Dhage on Unsplash

There is a social media post that circulates every now and then, sometimes titled “Rules for being human handed down from ancient Sanskrit.” 

Rule #4 states: “A lesson is repeated until it is learned.”

One of those lessons is being re-articulated by Claire Zau, who published a helpful overview of OpenAI’s “Study Mode.”

Near the end, she writes:

“Does [Study Mode] solve the cheating problem? No. But that feels more like an indictment of how we design assessments than of the tool itself.


“But I do think this is a meaningful step in the right direction. Like any tool, Study Mode is not a silver bullet. It’s only as effective as the pedagogical choices that surround it.” (emphases added)

Study Mode is not the story—ie, the lesson—here. As Claire points out, it is only a tool. 

She implies that the real lesson being repeated—because we have not yet learned it—is that we need to rethink assessments and instruction.

I would go even further.

We need a better mental model for assessments and instruction.

When I talk to edtech founders, I tell them all the same thing: You may have built the most effective AI [grader, lesson planner, etc.] in the world, but if it does not plug in to a mental model of powerful learning, then it will reproduce (and amplify) the current system.

At Middle States, we define powerful learning as an experience in which learners develop purpose and agency as they develop knowledge, skills, and dispositions. This definition is a loose paraphrasing of Mehta and Fine’s essential read, In Search of Deeper Learning (2019).

ChatGPT Study Mode is not fostering purpose or agency. And it is questionable whether it is fostering knowledge, skills, or dispositions. Which means that at best, tools like ChatGPT’s Study Mode, KhanMigo, Claude’s Learning are anti-cheating tools. 

But if we shift our mental model, maybe we can find a place for them to be tools that support powerful learning.

That mental model for me is games-scrimmages-practices. Let me know what you think!

Next
Next

The TAKE IT DOWN Act — And The Rise Of AI-Generated Harm