Gabrielle Likavec keeps up with research to guide her instruction as a teaching and learning consultant working with aspiring educators at Central Michigan University. The MEA member notes a few key points from early data emerging on AI.

> Within five months of ChatGPT’s release, 79% of surveyed C-suite executives reported at least occasional use of the technology, “so if we don’t prepare our students for using AI, we’re doing them a major disservice,” Likavec told attendees at MEA’s Higher Education conference in October.

> A study of coursework at 200 universities concluded in 80% of classrooms, generative AI could create a final product to be turned in with minimal human intervention. “That is what’s different; it produces products, meaning for us process has to become more of a focus.”

> Detection systems aren’t reliable. Numerous studies have shown at best programs accurately root out cheating students 60% of the time and at worst incorrectly flag original work from non-native English speakers, autistic people, and neurodiverse individuals, Likavec said.

“As students get better at writing prompts, that 60% number goes down. It’s an arms race, and there’s no way to get ahead of what these models and tools are doing, so we have to rely on other methods.”

For those reasons and more, AI is disrupting instructional practices. But the advancing technology also offers assistance to educators and benefits to students who make the shift, she says.

Likavec compares the potential for individualizing instruction to adaptive online tests in which students move through pathways depending on how they answer questions. “Now we have the same option not just for assessing learning but for helping students find a pathway to competency.”

In addition, chat bots can speed language acquisition for non-native speakers and accommodate students with learning disabilities. Neurodivergent students with executive dysfunction can be helped forward with organizing structures or step-by-step plans.

Despite her own initial reluctance, Likavec enjoys time-saving tools such as a rubric maker on Magic School AI. Nearly 70 offerings at Magic School, with K-12 and university levels, include tools to make AI-resistant assignments; give students feedback; and create a song on any topic.

Emphasizing process and relationships remains the best way for educators to discourage student cheating with AI, Likavec said. She encourages dialogue with students on how to use and cite AI and when to set it aside for their own good.

Assignments built around experiences further encourage students to deploy higher-order thinking in relation to content. Another adaptation asks students to reflect on learning, a metacognitive approach that is not new but takes on new importance in the age of AI, she added.

“One of the interesting things about artificial intelligence is it’s been trained to mimic human intelligence, so if you are aware of how you think, how you learn, how you accomplish a task with a goal and outputs, then you are better able to use AI. We know our students struggle with this.”

Educators can also take advantage of the fact that AI goes off topic, makes mistakes and adopts biases and stereotypes. “When using generative AI, our students should know to ask these questions: Is this accurate? Is it clear? Is it biased? Is it well-focused? Why do you say that? How do you know?”

A future without human beings evaluating AI outputs is frightening, which underscores the importance of educators’ work in this realm, Likavec said: “Without a lot of literature to draw upon, we have to do some trial and error, but that’s OK. We can do this. We have to.”

RELATED STORIES

Get ready—AI is transformative: ‘We have to focus on the speed of change’

Veteran: AI ‘democratizes education’

SVSU profs: AI can boost creativity

In daily crunch, AI is a time machine

ICYMI: Free AI resources for your district