Students will become adept at “truth-spotting”, judging what’s important, and ethics, with the growing use of generative AI, an educator has told a parliamentary committee.
Anthony England, director of learning technologies at Pymble Ladies’ College, said while GenAI tools are a challenge to education, they’re not as dangerous as is widely believed.
Speaking to the House of Representatives’ Standing Committee on Employment, Education and Training, England said GenAI needs to be understood not as an “answer machine” (like Google), but as a “making tool” that can, for example, produce text in response to prompts.
As a result, England told the committee, GenAI will change how educators view assessment tasks.
“Where teachers are really motivated to learn and invest time is in assessment practice,” he said.
The traditional essay – getting the student to make something – is intended as a way for students to “gauge their mastery of the content”, he said.
If something else can make the essay, “I’m not actually measuring the student, I’m measuring their access to generative AI,” England said.
So, he said, assessment needs to change: “The focus should be on the process, not the product – if the student is submitting an essay, there should be a footprint of how they got there.”
“A teacher should be able to check anyone’s progress at any time, and that audit will keep a lot of students on-task, and the notion that you’re keeping track of progress, and the process you used to get there, has got to be the focus on what makes a good assessment task.”
That, he said, puts learning design at the centre of how educators can respond to GenAI’s impact on education.
Rather than being assessed, he said, students need to “level up” and learn to be “copilots and assessors” of what they’re offered by GenAI.
Since GenAI is a maker of content, another skill students need is the ability to “look at what’s created by another product, and evaluate it.”
Bias
While bias in AI is a concern, England said, in GenAI bias is far less a problem compared to services like YouTube or TikTok which have “algorithms to send you down a clickbait rabbit hole” that over time becomes more biased and more extreme.
In comparison, GenAI’s reinforcement learning “will be quite circumspect and calibrate you back towards the centre”.
“I’m not as alarmed by the bias that is captured in the model, compared to other loud voices in our culture,” England said.
A challenge is to teach students to become “truth detectives”, he said, which is “is a really difficult skill … it’s hard to navigate truth in a fake world.”
Teaching good prompting
In practice, England said, good prompting is the basis for the effective use of GenAI in education.
Teachers “need to be given examples”, he said, “and then a place to apply those examples, in a way that supports their own goals.”
The teachers can then construct scenarios students can work through “where they have a chance to use the tool in response to guidelines they have been given.”
“To learn how to prompt well doesn’t take much time,” he said. “There’s just a few insights people need.”
Those insights are that they’re the boss; they need to own what “good” looks like; that they need to be quite descriptive in terms of what outcome they’re after; that guardrails need to be established; that they need to iterate and give feedback to the tools; and that they need autonomy.
“To summarise, freedom to use [GenAI] is great, but they also need to be given examples, and a context in which to apply those examples themselves.”
From there, he said, teachers will become “informed in their opinion” instead of playing catch-up.
Cheating and motivation
England said Pymble Ladies’ Colleges’ existing plagiarism policy is enough to cover the concern that students will simply hand over essay-writing to something like ChatGPT – and he believed that would be the case in most schools.
However, he added, educators need to look at what motivates students, because “if the only objective” of a task like an essay is a single “A, B or C” score, that’s “ripe for cheating”.
Assessment tasks will have to become “intrinsically motivating”, he said: “Each person wants to grow and be better in things that are important to them.
“That goes to the heart of what education wants to do for everybody.”