For as long as “educational technology” has been a discrete concept, discussions have ensued about how it might change our understanding of the purpose of formal education.
More recently, as artificial intelligence (AI) and machine learning models have become more sophisticated, even more attention has been given to these discussions.
While some current applications of AI are limited to narrow functions like grading and automating administrative tasks, the broader possibilities opened by such technology could alter our very conceptions of teaching and learning.
For example, as the Internet became integral to everyday life, a driving question in education arose: How do we assess learning when the answers are at everyone’s fingertips?
As artificial intelligence becomes more commonplace, this question might shift to ask: What does it mean to learn and produce knowledge?
It seemed a fitting exercise to ask two different AI programs directly how artificial intelligence might change K-12 education.
This exercise was helpful in determining how useful these tools currently are and in providing different ideas about how their use and impact might evolve in the future.
Below are the prompts I provided to the tools, snippets of the text produced, and some follow-up reflections of my own.
Asking the AI
Prompt #1: How will artificial intelligence change K-12 education?
Key Details from Response: …some potential changes that AI could bring about in K-12 education include improved personalized learning experiences for students, increased use of data-driven decision-making by educators, and greater use of AI-powered educational tools and resources by students and teachers.
While offering a few broad areas in which AI might impact education, this is obviously a pretty generic and uninformative answer. Still, it does provide somewhat of a springboard for additional research or questions to explore more deeply, possibly within the tool itself.
The AI’s response here tracks with the general impressions that some K-12 educators had when presented with this technology — its answer offers a good starting point but leaves a lot to be desired (or worried about, depending on your perspective).
To try to get more specificity and varied formatting, I tried a different approach:
Prompt: Write a 600-word blog post about the impact of Generative Pre-trained Transformer 3 on K-12 education
Some of the examples included in the response:
In October 2020, the district of Bellevue, WA announced that it would be using GPT-3 to help students with writing assignments…With GPT-3, the district will be able to use a much wider range of models, which will allow students to get more personalized help with their writing.
…a middle school teacher in New York City is using GPT-3 to create math lesson plans. By inputting a few facts about a topic, the teacher is able to generate a complete lesson plan, complete with worksheets and answer keys.
A company called Playtech is using GPT-3 to create an educational game about the history of the United States. The game will be able to generate historical events, based on input from the player.
These are all interesting anecdotes but are broad and lack a nuanced perspective that might be gained from real journalistic or research tasks like interviews and primary source analysis.
Additionally, with this and other prompts, the tools generated relevant information and data but provided no citations and, in some cases, needed more specificity to easily find sources for its claims.
A recent EdWeek piece offers a much more thorough analysis and leaves the reader with a keen understanding of the current impact that this specific technology is having on the field of education.
Some have speculated that the rise of technologies like GPT-3 and DALL-E will hasten the automation of creative work, contrary to the dominant theory that physical labor would be the first form of work replaced by machines powered by artificial intelligence.
While the foundations of teaching — like personal connection and relationship-building — may be impossible to automate, other educational fields, such as instructional design, may be ripe for disruption or rapid evolution thanks to artificial intelligence.
One could imagine a tool that produces comprehensive course scripts and syllabi, for example, or one that could be fed a topic, a content standard, and a desired output and present a polished learning activity. (I used OpenAI’s text completion tool to do just that and produced a writing activity, a rough rubric, and a writing sample.)
It could be the case that, like many other technologies throughout history, the impact of artificial intelligence and machine learning remains at the fringes of most people’s day-to-day lives.
It is clear that we are still in the very early stages of applying these technologies to common use cases, but by spending just a brief time with them, we begin to see the enormous potential for disruption.
At this stage, as James Thurber wrote, “It is better to know some of the questions than all of the answers.”
Anyone who holds a stake in the future of teaching and learning should be familiarizing themselves with these tools and formulating the right questions because as GPT-3 “personally” told me:
The full impact of artificial intelligence on education is not yet known.