Teachers: Could you use an extra 13 hours in your work week, or for your personal life?
That might be possible in the future, according to a report published last month by McKinsey & Company: “How Artificial Intelligence Will Impact K-12 Teachers.”
The report estimates that 20 percent to 40 percent of the tasks teachers spend time on—grading, lesson planning, general administration—could be outsourced to technology. But some teachers and AI experts caution that the report’s conclusion might be overly optimistic, and may apply to some groups of students and schools more than others.
The McKinsey authors acknowledge some limitations, but say that AI could help teachers prioritize the tasks that matter most. While AI—which is basically intelligent software that becomes more sophisticated in its thinking based on the amount and quality of data fed into it—can’t inspire students, resolve conflicts, or mentor and coach, technology could free teachers up to focus on those more important, and more rewarding, tasks, they argue.
Additional time could “help support social-emotional learning and the development of the 21st-century skills that will be necessary to thrive in an increasingly automated workplace,” the report says. “It will enable teachers to foster one-on-one relationships with students, encourage self-regulation and perseverance, and help students collaborate with each other.”
Or it could give educators extra time to focus on their personal lives, which in turn might make the profession more attractive, the report says.
One area that shouldn’t be outsourced, according to the report? The actual teaching. Widespread use of technology for direct instruction may actually do student learning more harm than good. The most recent Program for International Student Assessment found that students who used technology in the classroom performed worse than those who didn’t. The National Assessment for Educational Progress, or NAEP, cast similar doubts about the usefulness of technology. Not ‘One Size Fits All’
To be sure, automating teacher tasks comes with a whole host of logistical and ethical issues. For one thing, there are the serious privacy considerations with any technology that collects and analyzes student data.
Plus, Anastassia Loukina, a research scientist focusing on natural language processing and speech at the Educational Testing Service, said there are bound to be huge differences in just how much technology can help different types of teachers, in different types of schools, get their work done.
Technology, she said, may help teachers fill out forms faster. But the extent to which it helps with assessing student progress, lesson planning, and grading will depend a lot on the type of students the teacher is working with. A particular technology “may look like it works for everybody” but turn out not to work well with students in special education, English-language learners, or disadvantaged students.
Loukina expects that technology could eventually serve as a helpful partner for teachers, but it won’t be able to entirely do certain tasks, such as grade student essays. A machine, for instance, might be very good at checking whether a student used proper grammar in an English composition, but less skilled at connecting what a student wrote in a paper that was based on a recent class discussion.
“It’s likely that it’s not going to be one size fits all,” she said.
Teachers are also uncertain about the report’s conclusions.
Judith Stork, a 12th grade Spanish teacher in Columbia County, Ga., said she has some qualms about turning over grading to artificial intelligence, especially for student writing.
“I would be nervous [about letting AI] give feedback on a writing assignment,” she said. “It’s personal, it’s a conversation, you want authentic audience.” She expects students “would not give the same quality of output if they know it’s a machine reading versus a human.”
What’s more, she worries that affluent, well-resourced districts and schools would likely get access to the technology first, exacerbating the digital access gap between schools serving high percentages of disadvantaged students and those that don’t.
Still, at least some teachers do see the potential of AI to help with some of the time-consuming, but not necessarily student-centric, tasks. For instance, according to an Education Week survey conducted earlier this year, 44 percent of teachers thought AI could help in “taking attendance, making copies, and other administrative tasks.” Another 30 percent suggested grading and another 30 percent hoped technology could make it easier to communicate with English-language learners.
And some teachers are all in on the idea.
John Davenport, a 7th and 8th grade social studies teacher at Corte Madera School in Portola Valley, Calif., said he was surprised that the report estimated that technology could automate only 40 percent of a teacher’s job.
“I think mixing current ed tech with AI you could probably automate just about everything that teachers do,” he said, with two notable exceptions: mentoring students and managing their learning. “The face-to-face human interaction, I don’t think AI could replace that,” he said.
He would be eager, he said, to focus on that part of the job and outsource the rest. “It would be much more rewarding than putting grades into the student information system,” he said.
A version of this article appeared in the February 12, 2020 edition of Education Week as How Artificial Intelligence Might Save Time