For years we’ve been debating the effectiveness of homework.
Innovative schools point to the growing research that homework doesn’t benefit learning.
That failure is strange, though, because we know that learning requires what’s called “retrieval practice.” When you learn something new, it sticks much better if you remind yourself later in the day.
There must be a better way to do homework.
But, traditional schools kept assigning homework as-is, vaguely handwaving to principles like “discipline.”
But now Pandora’s box is open. There is no closing it. Kids can paste pretty much any question or assignment into chatGPT, and it spits out passable and hard-to-detect answers. It will write an essay, a complex math problem, or the correct answer to a multiple-choice question.
The age of traditional homework has come to a abrupt end.
Now we’re forced to seriously consider why homework wasn’t working.
Can we make it work for kids?
Why homework wasn’t working
If AI disrupted a teaching model, it wasn’t good to begin with.
Learning by coercion was never good. And even before chatGPT, kids were figuring out ways to cheat the system. Even if they weren’t cheating, they were going through the motions to get the grade, not to learn more about something that genuinely interests them. Only in rare cases, with great teachers, does subject matter “click” with a student.
Kids aren’t incentivized to learn. They are incentivized to trick teachers and expend minimum effort. That’s not a judgment on the kids; that’s actually the incentive structure–kids are “smart” to get the highest grade for the least effort.
If a student truly goes all out and learns something new, they can’t get higher than an “A,” so it feels like a slight punishment to spend the energy. Nobody wants to feel like a sucker.
Truly smart kids are best at gaming (manipulating the outcome) the whole schooling model. But, is that the best place to expend their brain power? In gaming the very system that was ostensibly built to foster their growth?
That’s why homework doesn’t work. The incentive structures are at odds with the theoretical goal of making kids learn. What actually motivates kids is a mixture of attention, relevant goals, and fun.
With the advent of chatGPT, that incentive structure’s flaw is exploited at the highest level. Now, not learning, but still giving the right answer, is cheaper than ever. Any smart kid will take that deal.
So, how do we align the goals of schooling to incentivize the kids to learn?
How to make homework WORK
If you want to be great at violin, you learn a new technique, practice it, put the violin down, and then practice it again later. That helps imprint the new skill onto long-term memory.
That’s the (theoretical) utility of homework.
The point of doing math isn’t to figure out what 2+2 equals for the benefit of the whole world–it’s to craft the mind of a person who can figure out what 2+2 equals.
Similarly, writing an essay is not to develop a world-shaking argument. It’s to craft the mind of someone who can think effectively. The essay crafts the writer as much as the writer crafts the essay. The benefits of composing thoughts into words are nearly endless: interviews, networking, self-discovery, and on and on.
But kids aren’t learning anything if homework is being thoughtlessly given to them without considering the goals, future, and particular interests of a specific child.
School must be more involved, more bespoke for each child, and stimulate more of an experience of play. That will make it more expensive, but that’s the reality of the world after chatGPT. Thankfully, school choice is increasing all around the country. We need alternative methods. Competition between many options will drive the cost back down again.
Practice (homework) should be scientific. Technology can solve this problem, and is already doing so.
The school choice movement will allow students to move to the schools that best fill their needs, rather than shoving busywork on them. With busywork now being hackable and cheatable, results in schools that lean on it will continue to plummet, while kids that go to schools with more room for customization will thrive.
The future really belongs to the people who rethink homework altogether.
What if we let ChatGPT become part of the homework? Instead of calling it cheating, what if we turned our assignments into things that involved the tool, and taught kids how to use it?
Of course, Learning does require a little discipline, but not because kids need to just shut up and do what they’re told. An attentive mentor who sees the potential in an individual student can help bring it to life. “Practice writing for 10 minutes tonight! Use chatGPT for research only,” is the encouragement from someone who knows and understands the kid, not a dictate handed down from the state. If they cheat, cheating themselves out of the growth experience, that’s on the mentor as much as the kid.
We know from research that long homework sessions are likely not helpful. Kids need time to run around and play–that’s part of their ability to be creative, think divergently, and synthesize everything they’ve absorbed in a day. The sweet spot seems to be about a 10-minute review of each topic they’re learning about.
So, to review, here’s how to make homework work:
Smaller, more attentive classroom settings
Using technology (like chatGPT and others) instead of fighting it
Teachers who are mentors
Educators who take responsibility for cheating as much as the kids
Shorter homework sessions–about 10 minutes
Make sure to follow us for more practical insights into how to deal with the rapidly changing world.
Thanks for reading,
Taylor + rebelEducator team
P.S.
What we’re looking at:
Quotes we’re pondering:
“Study hard what interests you the most in the most undisciplined, irreverent and original manner possible!” — Richard Feynman
“AI makes the value of knowledge go way down and the the value of wisdom go way up.” — Greg Isenberg
“One of the biggest misconceptions I had when I was a child was thinking that you had to understand something in order to do it.” — Simon Sarris
Thank you so much for the shout-out!
Point 2 ('integrating' ChatGTP)
smh -- are you f* kidding me?
If this is going to be a "this-is-fine" techno-utopian blog, count me out.