In various parts of the world, access to education is, or risks becoming, a huge crisis. UNESCO estimates that around 20 million new teachers are needed worldwide - and that's not taking into account the huge number set to retire in the next decade. With this level of demand, some specialists believe the key to filling the gap could be artificial intelligence.
AI teaching by 2027
Given, that subheading might seem farfetched. But Anthony Seldon, the British education expert who made the claim last September, stands by his prediction. It's widely agreed that robotization is on its way. But usually, we think of it in the context of jobs that are risky, monotonous, or otherwise unsuited to humans. In the context of education, many would object, human personality is important.
Teachers, after all, have a delicate task on their hands. They are responsible for a large part of children and teenagers' intellectual and emotional development. They have to communicate a wealth of knowledge to their charges, and at the same time ensure their behavior doesn't cause any problems. It's a job that, at first glance, seems to require a pretty non-robotic skillset.
Another expert, Rose Luckin of UCL, tones down Selden's claims somewhat. “I do not believe that any robot can fulfill the wide range of tasks that a human teacher completes on a daily basis," she says. But she remains optimistic about the possibilities for AI in education. The difference is that Luckin views AIs not as potential teachers, but rather as teaching assistants.
AI & Analytics
So what does Luckin suggest these AI helpers ought to do? A lot of things, it turns out. For a start, Luckin suggests, AI could certainly assist with the elements of teaching that are basically routine: Grading multiple-choice tests, taking attendance, finding appropriate lesson plans. But that's not all.
Luckin argues that if AI are suitable for routine, uncreative jobs, they are also suitable for roles involving data collection and analytics. She suggests outfitting future classrooms with various sensors - speech recognition software, eye-tracking, and so on - to monitor the behavior, mood, and attention of the class.
Teachers could use this data to determine which students need extra attention, either providing this attention themselves or directing the student to work with an AI-powered tutoring system programmed with that student's learning style in mind. Students could even access their own records to learn about how good and bad learning habits affect their performance.
The human element?
This risks being the most controversial part of classroom AI integration. Parents and students may object on principle to this sort of surveillance. And certainly, the sensitive data collected would need to be well-protected.
AI integration also risks drawing the ire of teachers who fear being automated out of their jobs. Those who consider AI potentially dangerous might also object. So might those who simply think teaching always requires a "human touch". But Luckin's proposals emphasize the secondary role of AI in teaching, as a tool like any other, and always subservient to the teacher.
There's still time to puzzle out these issues. Classroom AI integration likely won't happen for another decade or more. But if we want to make it the best we possibly can, it's important to think about it, like Seldon and Luckin, well in advance.
Now, would you trust your kid with a robot? Let us know in the comments below!