AI Tutors and Education Reform

2023 March 21 Twitter Substack See all posts

AI can revolutionize K-12, but how many actually want it?

The Inconvenient Lack of Demand for Higher Quality Education

Despite many lamenting the state of K-12 education in America, almost three-quarters of parents are satisfied with their oldest child's education. Unfortunately, we don't know why the remaining quarter is unhappy. Is it due to a lack of friends, a specific teacher, or their child's poor performance on NAEP tests?

Parents are mostly happy with their children's education. Gallup

As someone who greatly benefited from school choice and ambitious curricula, it has taken me years to internalize that most parents in the US already have options with which they are content.

What Drives Apathy?

My favored model is that schools have an iron triangle:

  1. Parents need someone to watch their kids while they work. They often don't want to think about it, and they are usually content as long as their kids can reach the same status or skill levels they have.

  2. Most students prefer doing as little work as possible and focusing on fun activities like sports.

  3. Teachers don't want to be bothered by disruptive students, nagging parents, or overbearing administrators.

Teachers provide the minimum instruction required to meet parental expectations. Students can avoid doing most work by showing up and behaving. Parents don't bother teachers as long as their children come home intact. Teachers don't bother parents as long as their kids behave. They also organize to limit administrator influence, often allying with students and parents. Honors programs provide a relief valve for teachers, parents, and students with more ambition or stronger tendencies to seek status. The most ambitious still need to seek outside tutoring.

Everyone is happy, and these stakeholders might falsify records if the government demands "higher standards." They will fight hard against binding requirements or innovations like teaching scripts that challenge this compact.

The Collision of Digital Systems with the Triangle

The number two pencil is one of the most critical tools for our iron triangle. Grade books need adjustments at the end of semesters to pass on underperforming students or raise a grade for an unhappy student (or parent). Sometimes state-mandated standardized tests need altering as well. Teachers can make these changes using an eraser or the keyboard on an unconnected spreadsheet without being charged with fraud.

Many education reformers dream of systems where students use personalized software. There are some schools where it works wonderfully. One of my high school mentors joined a non-profit education organization to assist rural schools wanting to adopt these systems. 5-10 districts adopted the program, but many encountered issues. Teachers, parents, and students revolted at one of the most successful implementations, forcing the district to revert to the old ways. Another became a crime scene because teachers didn't realize that a central database tracked the changes they made to student grades at the end of the semester.

AI Chatbots and Education Revolution?

Advancements in AI, like ChatGPT, promise to become incredible educational tools. However, they will have a mixed impact if there isn't widespread demand from students, parents, and teachers. Any implementation connected to central databases will likely face tremendous pushback. ChatGPT-for-schools must be compatible with a "Gentleman's C" for widespread adoption. It could even be popular if it helps with classroom control, allows students to goof off, and lets parents believe their students have world-class teachers.

Of course, motivated students will be getting a world-class education! The gaps between the top and bottom rungs could grow to extreme sizes. As Tyler Cowen says, average is over.

Many hope bots will increase motivation for students, making them more ambitious. It would be nice if that were true, but schoolwork will likely still be dull compared to the perfectly customized entertainment these programs will be capable of generating.

Business to the Rescue?

Businesses already spend significant amounts of time and money on worker training. Rapidly falling costs for education should dramatically increase the volume they supply. McDonald's and other employers must adapt processes to workers with poor reading, math, and communication skills. Employers might save money and reduce the complexity of their systems by using a chatbot teacher to upskill employees for an hour per shift. The numbers could make sense for workplaces to teach "remedial" skills alongside job skills. Better pay and job duties incentivize workers to learn. The lowest-paid workers are the cheapest to train as costs fall because they have the lowest opportunity cost.

There are diminishing returns to keeping unmotivated students in institutions that resist external standards, especially when costs are falling for alternative learning models. Students could move on to working or further education before they turn 18. Many of the most capable and desirable students could be ready for college-level material years early with the help of AI tutors. It is certainly possible that college track kids will increase leisure or learn less valuable things while waiting to graduate. However, there is an opportunity for educational institutions that can meet the needs of these students. I'm not optimistic that we will maximize our new AI potential by increasing the size of the workforce and shortening adolescence, but it is a fascinating idea.

The Impact of AI Tutors

It seems like a slam dunk that ambitious students will race ahead, and businesses will increase training volume. Any impacts on the K-12 system are much less clear. Most parents and students are happy with the current system and already tolerate wide disparities in achievement. I certainly won't hold my breath!