The urgent need to overcome the hurdles faced by challenged learners has been a driving force in my 30-year career in education.
There is a personal component to this calling. My grandfather never went to high school and my dad never graduated from high school. We were working class poor, with the emphasis on poor. My wife faced even greater challenges. She didn’t speak a word of English when she immigrated with her family to the U.S. and was thrown into a sink-or-swim classroom. Her language challenge was exacerbated by the fact that she was mildly dyslexic, a condition identified much later in life.
My obsession with generative AI models such as ChatGPT and Bard lies in part with their highly anticipated ability to personalize learning for every child and meet their diverse needs.
A Tutor for Every Child
The Khan Academy launched a pilot for Khanmigo, its AI powered learning tool, in April of 2023. In the fall of 2023 that program went nationwide. Because Khanmingo is powered by ChatGPT 4 it is not free. The initial price of $60 per student per year has been slashed to $35 per student/per year.
Hopes are high and the rhetoric matches. According to Sal Khan, the founder of the organization, it will “ enable every student in the United States, and eventually on the planet, to effectively have a world-class personal tutor.”
According to a recent New York Times article, “Simulated tutors (such as Khanmigo) could make it easier for many self-directed students to hone their skills, delve deeper into topics that interest them or tackle new subjects at their own pace.”
Indeed, the possibility that every learner could have a personalized tutor thanks to gen AI is one of the great promises of the technology. We will for the moment put aside our concerns with the phrase “self-directed students.”
Early Research Offers Some Surprises
As you can imagine, generative AI is so new to the K12 workspace that there isn’t much research available. One has to examine research from higher education or the corporate sector to get a preview of what might occur in the elementary, middle and high school classroom.
A fascinating study by the University of Minnesota Law School was published earlier this year. The positive impact of study assistance provided by ChatGPT 4 “depended heavily on the student’s starting skill level; students at the bottom of the class saw huge performance gains with AI assistance, while students at the top of the class saw performance declines.”
An unpublished 2023 study from MIT asked college-educated professionals to complete mid-level writing tasks. Remarkably, “inequality between workers decreases, as ChatGPT compresses the productivity distribution by benefiting low-ability workers more.”
So what do these early studies mean for learners who face academic challenges? Will access to ChatGPT level the playing field or allow struggling students to gain ground on their more accomplished peers? Will it impede the highly skilled or academically gifted, a component of any diverse classroom?
Using ChatGPT to Personalize Learning
As much as I like to read research or policy, practical application is my comfort zone. As noted here before, I have spent the last eight months completing training programs in prompt engineering. That is the fancy term for telling ChatGPT what to do.
I left the classroom in June of 2023 but I have hard-coded memories of the varied reading levels of my sixth-grade students. I have been testing ChatGPT 4’s ability to design a differentiated curriculum and came up with the following prompt:
“I need you to create a 200-word reading passage about the states of matter. I want three versions of this passage: The first written at a fourth-grade reading level, the second at a third-grade reading level, and the third written at a second-grade reading level. When I approve the final versions of these reading passages in English I would like you to ask me if I would like the passages translated into any additional languages.”
ChatGPT easily completed the task. (Note: The keen reader will ask why I didn’t use a more technical term such as “Lexile level.” I can’t, because that would violate ChatGPT’s copyright rules.)
Differentiated lesson plans were next on my chore list for Chat GPT:
“I need you to create a 45-minute engaging, interactive introductory lesson for my sixth grade science class. The topic we are studying is states of matter. This lesson plan should take into account various learning styles. I have a large group of visual learners and a large group of kinesthetic learners.”
Again, it ably complied.
How to use Gen AI to Increase Personalization
On Feb. 3, the staff of New Tech High Center for Excellence will be offering a series of workshops on generative AI at the Napa County Learning Innovation Summit. Following a keynote by author/superintendent Michael McDowell, our staff will provide K-12 educators and administrators with insight on AI tools, techniques and strategies, with personalization being a throughline. This will also mark the launch of our AI-focused professional development.
We hope to see you there. In the meantime, send me your ideas for differentiation using gen AI. The hurdles are still there, and they haven’t got any lower.
David Ross (@davidPBLross) is the retired CEO of the Partnership for 21st Century Learning and the former Senior Director of the Buck Institute for Education (now PBLWorks). David was an 11th grade American Studies (History and English 11) team teacher. David created curriculum design templates, exemplary projects, rubrics for critical thinking and collaboration, and project management techniques.
