Last week, my second daughter vented about “baby doctors.” That’s not unusual. What was different this time was her claim that AI is making them dumber.
My daughter is a nurse at a major teaching hospital. At the ripe old age of 28, she is already an experienced hand at wrangling hospital residents—new doctors with fresh diplomas and little real world experience. She calls them “baby doctors.”
New residents make a lot of mistakes. That’s not their fault, she said, because they don’t know anything yet. Usually, however, over the course of months, they figure out why they were making mistakes and how to make correct decisions.
This last group? Not so much, she said.
“I know they are using AI,” she said, and they fixated on what the algorithm tells them instead of looking at the big picture of things affecting their patients’ health.
Yes, her account is anecdotal and it probably is biased. Nurses don’t have to work in a hospital very long before they lose respect for doctors in general.
But, her opinions touch on the hot topic of the role of AI in education. AI, by the way, is short for “artificial intelligence,” although the appropriateness of the term “intelligence” is quite controversial.
AI is the latest in a long line of technologies what were supposed to modernize and revolutionize education by separating students from teachers. The debate about the role of AI in teaching and learning repeats the cycle of hype and concern that attended the introduction of each of those technologies since 1950.
Television, which was supposed to open education up to a broader audience by freeing learning from the confines of the traditional classroom, was the first such technology to promise to modernize education following World War II. The hype lasted a few years before it became apparent that TV could not replace the vibrancy and interaction of live education, and that well-done educational TV was very expensive. Eventually, opinion settled on accepting that educational TV could play a supplemental role in learning, but could not provide the foundation and structure that learners needed.
Videotapes were supposed to improve on educational TV by giving learners more control of their pace of study and of any needed repetition, but still suffered from the lack of iterative responses to learner questions and comprehension that takes place in face-to-face instruction.
Computerized learning added to instant grading of exercises to the ability to repetitively watch videos, but still lacked student-teacher interaction.
Each of these was hyped as a revolution in education. In each case, some self-motivated learners did well with the new technology, while most learners struggled with it.
“Virtual charter schools” grew to prominence in the twenty-teens as a way to pair video and computerized learning with scheduled sessions with a live, but distant, instructor to again “revolutionize” and improve education. Such schools were permitted in 17 states by 2015. However, the National Study of Online Charter Schools, published in November 2015, showed that the performance in math and reading by students in virtual charter schools was much weaker than the math and reading performance of similar students in conventional schools. The authors suggested that it was harder for online schools to focus students on their work than for teachers to do so in person, which was one of the major drawbacks shared by television, videotape, and computerized classes. The study was conducted by researchers from the Stanford University, the University of Washington, and Mathematica Policy Research.
Nevertheless, when COVID-19 swept around the world in 2020, many school systems rushed to convert to virtual learning. The results were not good.
Worse outcomes from virtual learning were reported around the globe. A systematic review of studies comparing learning and school performance with virtual instruction to face-to-face instruction found that two-thirds of the studies reported worse outcomes from virtual learning; one-sixth reported stable performance; and one-sixth reported improvements. The reviewed studies came from Europe, Asia, North America, and South America.
A separate track has been to bring the technology of virtual learning, i.e., electronic devices and the internet, into the in-person classroom to combine the benefits of the volume of information on the internet with live interactions between learners and teachers.
Sweden, which was an early leader in incorporating electronic devices in school-based learning, re-evaluated that strategy in 2024. The country decided that there are real educational benefits to reading printed books and writing by hand and reincorporated those elements into schooling. They didn’t abandon digital devices, they simply got past the hype that devices can do everything and considered that their role in learning is limited.
What will be the role for AI? I think we’re still in the hype stage and have yet to figure that out.
Schools, especially public schools, have to balance many often conflicting goals, and how they strike that balance will affect the role they ultimately provide for AI. They are supposed to develop the individual to help students reach their potential and train workers for business and industry, which are two goals that may overlap but are not identical. They are supposed to teach students of similar ages who are at different levels at the same time so as to keep slower students progressing and advanced students moving ahead.
Whatever role AI finds in education, students will have to learn how to evaluate AI answers to determine whether they are good answers. Critical thinking, according to my daughter, is what the newest crop of hospital residents have yet to learn.
Resources:
While we debate the term “intelligence” in respect to AI, perhaps we should note it’s “artificial “