Bears Repeating from Jess Hill: Thinking critically about the role of AI in education
“The moon's an arrant thief, And her pale fire she snatches from the sun.”
― William Shakespeare, "Timon of Athens"
If you are in the field of education and preoccupied with teaching students to think for themselves — or in any profession with a healthy skepticism surrounding the automation of your industry due to artificial intelligence — you probably have been swamped with discussions and articles about “intelligent” computer systems. The current conversation seems to contain two schools of thought. One view is that ChatGPT or any form of generative artificial intelligence will cause the end of critical thinking, or, let’s face it, the end of civilization as we know it. The other view, equally distorted, is that AI, anthropomorphized by Hollywood and sci-fi movies, will solve our problems and make life and work much more efficient.
So where does that leave us as parents and educators who want our children and students to learn to think for themselves (not via bot) but also keep pace in a tech-driven future? Before going any further, I confess that I am not a technology or AI expert in any way. I have read quite a few articles about it and taken part as our faculty discusses the pros and cons of AI in teaching and learning. Thus far, in this nascent stage of LLM (large language model) technology, I tend to end up where I usually do, somewhere between the extremes of gloom and doom and Star Trek.
The reality is, AI is everywhere, and I do not want to underestimate the quakey feeling that a radically new technology elicits. In the 1970s, the first articles about artificial intelligence predicted that there would be no need for teachers by the end of that decade — AI would take their place. I am sure the fear in teacher lounges across America was exponential. On a smaller scale, I remember when graphing calculators were first introduced in the math classroom. Would there be a need to solve problems, learn calculus, use probability and statistics, or even graph a line ever again? We all wanted to know the answer, so the best teachers I knew leaned in to learn more about the technology alongside their students. They discovered the best ways to use a calculator as a tool for learning but not as the singular answer to every question.
This approach mirrors how our teachers are engaging with AI in this iteration. While more prominent and more powerful than any previous model, AI does not signify the end of in-person teaching. It actually calls on us to more fully acknowledge the value of creative thought, nuanced discussion, and personalized support in educating and mentoring children and young adults. Good teaching is not just about imparting knowledge but also about inspiring growth and curiosity.
Dr. Chris Dede, a researcher and professor at Harvard Graduate School of Education, has written blogs and articles that have helped my understanding of the potential strengths and limitations of generative artificial intelligence.
He compared AI and the new LLMs to a “digital parrot” because they can recall and give back whatever information they have cached. No doubt, generative AI is a vast storehouse for language and words from all cultures and contexts — far beyond the capacity of a parrot. It has conquered our use of idioms and the informal ways we seek knowledge and communicate. It doesn’t take much imagination to begin to think that artificial intelligence has human qualities, but make no mistake, it does not. Dr. Dede states that AI is a “brain without a mind.” Any human being, educated or not, is able to have feelings and a sense of moral and ethical judgment. AI does not. It is not able to plan or think the way we do. Nor can it reevaluate a conclusion after new understanding and broader experiences are gained. Intellectual humility is far beyond its scope.
Even with these limitations, machine intelligence cannot be completely written off in education. Generative AI is able to make accurate computations and predictions based on those calculations. It is good at reckoning and natural language processing. It can enable self-directed learning outside of the classroom. It is not something to fear but rather something to understand so we can find ways to harness it and have it best serve us. As with any tool, we can use it to do things that weren’t possible before.
That brings me to perhaps Dr. Dede’s most profound conclusion. He uses the words of Shakespeare (which were definitely not written by AI) to make his final point. “The moon's an arrant thief,” he quotes, “And her pale fire she snatches from the sun.”
Although dramatic in language, Dr. Dede’s point is that, in the case of artificial intelligence, the moon (AI) reflects the sun’s light (human knowledge), and the world has always worked that way. Books reflect our knowledge. Documentaries reflect our knowledge. Podcasts reflect our knowledge. Now, AI reflects our knowledge. Education, he says, can evolve to work with — rather than fight against — artificial intelligence.
Where we must be cautious, Dr. Dede adds, is in situations where AI steals light from itself, compounding the chance that it reiterates erroneous information as fact or removes unique human thoughts from the equation. The worry is: “We are treating moonlight as if it is sunlight,” Dr. Dede says. “We are starting to dump large amounts of generative AI material on the World Wide Web. If we are now feeding moonlight into the sunlight, then we are dumbing down the sunlight.”
Our job is to help our students distinguish the real from the stolen light.
Earlier this month, I was in a middle school faculty meeting as the teachers discussed the best ways to help middle school students evaluate online information. How can they investigate their sources and find the author’s or creator’s purpose? How can they learn to track down the original source? Surely, this is the beginning of the education they will need down the road in deciphering original writing and thought from the curated version.
As I decipher my own learning on the topic of artificial intelligence, here is what I have come to believe: If our highest form of assessing intelligence is a standardized test, then AI will win almost every time. But if we acknowledge that experiential and collaborative learning and formative assessment are the most effective and engaging forms of teaching, learning, and assessing, then we will develop capabilities in our students that AI can’t touch. We will develop students with good judgment and empathy and a nuanced understanding of problem-solving in unique real-world circumstances. Our students will also find the best teachers to be human and not large language models. Teachers remember what it feels like to learn something new and what it feels like to begin a new school year with all of the associated emotions of those first weeks. We all have that over AI.