If AI teachers ever learn to care about students, that will be the least of our worries

Tánaiste Michéal Martin hoisted a kite this week about reducing the Professional Masters In Education (PME – post-primary teaching qualification) from two years to one.

The week before, Norma Foley announced that she was postponing a proposal that teachers assess their students as part of the Senior Cycle reform. She cited the advent of generative AI as the reason why the State Examinations Commission should continue its current practice of overseeing and externally examining projects and coursework.

Of course, the Tánaiste’s proposal is sensible. A four-year degree followed by a two year PME is a long, expensive process. Something that eases that financial burden for would-be teachers while simultaneously helping with the recruitment crisis in schools has to be a winner.

No proposal will please everyone, not least universities receiving two years’ fees at the moment. Not to mention the poor devils who paid those fees for two years and who will perhaps see colleagues who only did a one-year PME sailing past them into jobs.

Did no alarm bell tinkle from RTÉ's editorial structures about its programme on abortion services?

---

Critics are right to worry that hate speech laws are open to abuse

---

Leaving Cert reform plans may result in irreparable damage to exams’ reputation

---

No smartphones before 14: Is this the prescription for a happy childhood?

---

Nonetheless, there is a bigger question than the length of the PME, which is what the future holds for teaching in general. It is not unrelated to Norma Foley’s avowed reason for postponing the assessment of coursework by teachers.

Education has had many faces but it has always had a human face. Is that about to change? Will students interact primarily with generative AI in the future?

Some people believe that generative AI or large language models will herald a revolution that will make Gutenberg look insignificant. Specifically, there is a belief that education will be completely transformed.

Most people’s first thought about AI was how easy it would make cheating. Another equally strong faction believes it can become an invaluable assistant to teachers. For example, it can generate lesson plans, exit tickets (activities that test learning at the end of a class) and rapid assessments. By speeding up these routine activities, it will allow more time for teachers to interact with students.

[ Artificial intelligence: We need to be ready for rapid evolution ]

[ In AI we trust... or do we? ]

Then there are those who believe that it will slowly erode the need for human teachers or even the present model of schooling, which involves an adult in a classroom with 30 or so students, all of whom are roughly the same age.

‘Industrial’ model

This is often referred to as the “factory” or “industrial” model of education, named for the time when learning allegedly moved out of the home and into large public institutions. It’s not so simple. For example, before that model, the monitorial system existed in which hundreds of students were divided into groups, each in the charge of a monitor who was usually an older student. All the groups were under the ultimate control of one, presumably superhuman, teacher.

Education has had many faces but it has always had a human face. Is that about to change? Will students interact primarily with generative AI in the future?

Salman Khan, founder of the Khan Academy, believes that gen AI heralds a true revolution. He founded Khanmigo by partnering with OpenAI, which developed both ChatGPT and DALL-E 2, an AI art generator. Khanmigo is an AI tutorial system which involves a chatbot prompting the student to work things out for her or himself, rather than just supplying answers.

Cannily, if you will pardon the pun, Khan has marketed Khanmigo simultaneously to teachers and students.

For teachers, it promises to reduce workload. For students, the AI-powered chatbot claims to help students develop critical thinking.

Newark’s public schools have enrolled in a pilot program using Khanmigo. The reviews were mixed. The AI tutor gave too much help too soon. But the whole point of the exercise is to train the AI to work better and no doubt, Khanmigo is being constantly tweaked and improved.

There are good reasons to be sceptical about the next alleged revolution in education. Derek Muller, a superb science educator working online, has a 2014 video, The Most Persistent Myth, that amusingly makes this point. He goes through all the innovations that were allegedly going to cause an educational revolution. For example, in the 1920s, Thomas Edison declared that the motion picture would “supplant largely, if not entirely, the use of textbooks.” Muller does not mention Sal Khan’s 2011 Ted Talk with Bill Gates about how video would revolutionise education, but he could have.

Derek Muller says: ‘The most important thing a teacher does is make every student feel important and make them accountable for the work of learning.’ That still happens best in a social environment with a caring teacher

But is this one different? Is AI the real revolution? Despite the fact that AI suffers hallucinations (making stuff up when it does not know the answer), it will eventually become more accurate. Or even, like the best teachers, learn to state, “Great question to which I do not have a great answer so let’s try to find one.”

Caring teacher

Even though Muller’s vlog long predates ChatGPT, I think his conclusion is still accurate. Teaching is about more than transferring knowledge.

Muller says: “The most important thing a teacher does is make every student feel important and make them accountable for the work of learning.” That still happens best in a social environment with a caring teacher.

At the moment, all generative AI can do is give a convincing facsimile of a caring teacher, and it cannot demand accountability. If it ever genuinely learns to care or make students accountable for their own learning, then humanity will have a much bigger dilemma than its role in education.