In teaching patience is a virtue...

But only up to a point.

Digital tutors are here. We have no idea if they “work” both because they are too new and perhaps more importantly, because we have not bothered to discuss and agree on what “work” means in this context.

This strikes me as a problem. It’s possible that digital tutors work great, except that what they work at is not something we actually want students to do.

The undeniable efficacy of these digital tutors is the book-length claim of Khan Academy and Khanmigo founder Salman Khan’s new infomercial (oops…book), the truly bizarrely named, Brave New Words: How AI Will Revolutionize Education (And Why That’s a Good Thing).

What Is the Purpose of Education?

There is actually significant evidence that these digital tutors will not “work,” even when measured against the claims of those who believe in them most strongly, primarily because similar experiments have been tried many times before, always running into the same problem well articulated by Satya Nitta, an IBM researcher who worked on creating a digital tutor based in the company’s Watson platform: “We missed something important. At the heart of education, at the heart of any learning, is engagement.”

Given the title of this newsletter, the problem of engagement is obviously close to my concerns, and I intend to write in more depth in the future about the many reasons why I believe outsourcing instruction to digital tutors (particularly for writing) is a bad idea,¹ but today I want to talk about one of the purported advantages of these tutors, their “patience.”

In Brave New Words Khan describes Khanmigo as “a personalized and patient tutor that focuses on the students’ interests or struggles and empowers educators to better understand how they can fully support their learners.”

In a blog post at his own website, investor and self-declared “tech-accelerationist”Marc Andreessen went even further on the patience front, saying that the AI-powered tutor would be infinitely patient, along with infinitely some other things as well:

Every child will have an AI tutor that is infinitely patient, infinitely compassionate, infinitely knowledgeable, infinitely helpful. The AI tutor will be by each child’s side every step of their development, helping them maximize their potential with the machine version of infinite love.

I want to question the desirability of this infinite patience when it comes to teaching and learning contexts, and suggest that the uncritical selling of this trait as a de facto positive by boosters of generative AI is just one indicator that they have not spent all that much time about the underlying dynamics of a teacher-student relationship.

Honestly, I don’t think they consider relationships as part of learning at all, which is a problem not unique to this cohort, but we have reams of evidence that positive student/teacher relationships have significant benefit to both the quality of teaching, and the amount of student learning.

Seminal research as part of the Gallup-Purdue Index showed that how professors relate to their students has a significant effect on how “engaged” (there’s that word again) they are at their jobs:

Being engaged at one’s work is especially important because it significantly increases one’s likelihood of experiencing overall well-being.

Essentially, if students had experienced some form of faculty mentorship in college, they were more likely to be living a happy life:

I had a mentor who encouraged me to pursue my goals and dreams.

I had at least one professor who made me excited about learning.

My professors cared about me as a person.

What is striking to me is how low the bar for this form of mentorship is and at the same time how it is definitionally impossible for an AI tutor to achieve even this low bar.²

So, we know that relationships matter to students. We know that to the extent a digital tutor could forge a relationship with a student, it would be an illusion (or even a delusion) the student is required to maintain for themselves. Now, let’s examine whether or not infinite patience is a desirable trait in a tutor, be they digital or otherwise.

As I understand it, the theory of the desirability of infinite patience rests on the belief that no matter how long it takes, the digital tutor will not fatigue or grow exasperated with a student’s wrong answers. This is a benefit because of course students have infinite patience for banging their heads against a problem that to them seems unsolvable.

Also, students are always motivated to continue to engage with the digital tutor no matter how long the session has gone on.

And let’s not forget that students are definitely always prepared for learning and come ready to give their best efforts no matter the current conditions or their level of interest, asking the kinds of questions that would unlock the right answers that allow them to learn.

And finally, a student would never provide an answer accepted by the tutor without having sufficient understanding of the meaning of that answer, or how that answer may connect to other aspects of the subject at hand.

The theory of the benefits of an infinitely patient tutor suggests that patience is the most important trait of a good, teacher who is capable of motivating student learning. I think this is wrong. For sure, teachers need to have at least some patience, but is an infinite amount of patience truly a benefit to student development?

I recall many excellent teachers from my past who knew how to use their impatience strategically, like my 8th grade language arts teacher Mrs. Thompson’s arched eyebrow, employed when we collectively seemed to be off task. That arched eyebrow was effective because Mrs. Thompson had established a relationship with her students that was a mixture of respect and liking, mixed with just a soupçon of fear, that was primarily rooted in her reputation, and (likely invented) stories handed down like folklore from year to year.

When I have worked with students and they appear rather lost, sometimes the right response is patience - working backwards to see where we can reconnect - and sometimes it is impatience - as when it becomes apparent that the expected and necessary preparation has not been done. Once I hear the reason for the lack of preparation, my patience may return, or maybe I will pivot to some other emotion appropriate to motivating the student.

Something that exhibits infinite patience is likely to come in for significant abuse as it appears willing to absorb whatever indignities are thrown its way. I wish this were not the case, but I am also well-acquainted with human nature.

One of the first things people did upon the introduction of Siri’s voice assistant is swear at it, which would elicit a mild chastisement. ChatGPT seems to ignore any direct abuse you throw its way, which is probably a wise move, programming-wise, but imagining the kind of behavior I’ve directed at it (purely as an experiment) being targeted at a teacher is almost unthinkable.

Maybe digital tutors will be more useful than some of us think, but the idea that its infinite patience is actually a desirable trait in a teacher requires us to ignore our real-world experiences and knowledge of how teaching works.

1. If you want some thoughtful work on why these digital tutors won’t work, primarily coming from a math context, I recommend

Dan Meyer

s’ Mathworlds newsletter, particularly this post responding to some of Marc Andreessen’s claims.

2. Distressingly, the Purdue-Gallup Index also found that relatively small proportions of students reported having had these experiences in college, which is a problem in and of itself.