Educators Technology

If you are working on designing course-level AI policies, this paper I published in Innovations in Learning may be useful.

In the paper, I argue for the importance of course-level AI use policies in higher education. Institutional AI guidelines are important, but they often remain too broad for the day-to-day realities of teaching.

Faculty still need to decide what AI use is allowed in a specific course, how students should disclose it, what counts as misuse, and how AI fits with the learning goals of the course.

I reviewed literature from AI ethics, governance, learning sciences, academic integrity, and assessment theory, and I also looked at emerging policy guidance from international organizations and U.S. state-level frameworks.

The main argument: AI policy should not stop at the institutional level. It needs to reach the syllabus, the assignment sheet, the classroom discussion, and the assessment design.

The paper offers design principles for course-level AI policies, including:
collaborative development and student voice
alignment with institutional policies
pedagogical coherence
transparency
ethical modelling
equity and access
accountability
regular review and adaptation

I also provide a sample course-level AI use policy that faculty can adapt to their own disciplinary and institutional contexts.

My hope is that the paper helps instructors move from broad principles to practical classroom guidance.

Reference:
Kharbach, M. (2026). Developing course-level AI use policies in higher education: From principles to practice. Innovations in Learning, 1(2).

This is one of the foundational AI literacy papers I keep recommending for researchers working on AI and education:

Long and Magerko’s “What is AI Literacy? Competencies and Design Considerations.”

The paper outlines 17 competencies, including recognizing AI, understanding narrow versus general AI, identifying AI’s strengths and limits, understanding the role of data, recognizing human involvement in AI systems, and examining ethical issues such as bias, transparency, privacy, and accountability.

It is especially useful because it gives researchers a structured way to think about AI literacy before jumping into curriculum design, policy, or tool use.

For anyone studying AI literacy, AI in education, or digital competence, this paper is still a must-read.

Reference:
Long, D., & Magerko, B. (2020). What is AI literacy? Competencies and design considerations. Proceedings of the 2020 CHI Conference on Human Factors in Computing Systems, 1–16.

Using AI with children is a controversial topic, and for good reason.

Children are still developing emotionally, socially, and cognitively. They are forming values, learning how to reason, and building the metacognitive skills they need to judge information. Many are not yet ready to evaluate AI outputs, detect hallucinations, recognize bias, or know when a tool is quietly shaping their thinking.

That is why AI literacy with children should never be reduced to “teach them prompts” or “let them use ChatGPT.”

This handbook, Teach AI Literacy: A Guide for Teachers, is especially relevant for teachers in upper primary and secondary settings.

One important point from the guide: GenAI should not be used when children lack the metacognitive skills to fact-check and critically evaluate its output. It should also be avoided when learning depends on emotional development, values formation, foundational conceptual understanding, physical activity, play, outdoor learning, or human relationships.

This is the kind of AI guidance schools need: cautious, practical, child-centered, and grounded in learning.

Link in the first comment!

AI can help students generate ideas, but it can also make everyone sound strangely similar!

One of my favourite classroom activities is simple.

I give students a topic and ask them to work in small groups. Each group uses AI, usually ChatGPT, to help with brainstorming and ideation.

Then we come back as a whole class and compare the ideas.

What students usually notice is fascinating: the same ideas appear across groups, only dressed in different words.

At first, they find it funny. Then they start to see the deeper issue.

AI can make ideation faster. It can produce many suggestions. It can help students move past the blank page. But when several groups use the same tool, with similar prompts, the class often ends up with similar thinking.

This study by Anderson, Shah, and Kreminski confirms what I have been seeing in these activities.

The researchers compared ChatGPT with Oblique Strategies, a non-AI creativity support tool. Participants using ChatGPT generated more ideas, ideas across more categories, and ideas with greater detail.

So yes, ChatGPT helped.

But there was a problem. At the group level, ChatGPT users produced ideas that were less semantically diverse. Different people using ChatGPT moved toward similar ideas.

That finding is important for teachers.

If students begin with AI, they may get quick ideas, but many of those ideas may already be the obvious ones the system gives everyone else.

The sequence I now prefer is simple:

Students first generate their own ideas.

Then they use AI to stretch, test, refine, or challenge those ideas.

Then they compare what changed.

That small shift protects ownership. It also helps students see AI as a thinking partner, not a replacement for their own imagination.

AI can support creativity, but students need to bring something of their own to the table first.

Link in the first comment!



References

Anderson, B. R., Shah, J. H., & Kreminski, M. (2024). Homogenization effects of large language models on human creative ideation. In Creativity and Cognition (C&C ’24), 413–425. Association for Computing Machinery.

Here is another interesting research paper on creativity in the age of AI.

In this paper titled On the Creativity of Large Language Models, Franceschelli and Musolesi’s ask a deceptively simple question:

Can large language models really be considered creative?

Their answer is nuanced.

LLMs can certainly produce texts that look creative. They can write poems, stories, essays, scripts, and metaphors that many readers find impressive. But the authors argue that producing creative-looking artifacts is not the same as being creative in the deeper human sense.

Using Margaret Boden’s framework, they examine creativity through three dimensions:

Value: LLM outputs can be useful, attractive, and high quality.

Novelty: LLMs can generate texts that are new to the user and may not directly copy from training data.

Surprise: this is where things become more difficult. Current LLMs are trained to produce probable continuations based on existing patterns. This makes them good at recombining familiar ideas, but much weaker at transforming a field or creating genuinely new styles of thinking.

The distinction I found especially helpful is between creative products and creative systems.

An AI-generated poem may feel creative to the reader, but the model does not have intention, intrinsic motivation, lived experience, self-awareness, or a personal reason for creating it. It does not decide to write because something matters to it. It generates text because it has been prompted to continue a pattern.

The paper also uses the classic four Ps of creativity:

Product
Process
Press
Person

LLMs may perform well at the product level because the outputs can be impressive. But they are much weaker when we look at process, social context, and personhood. They do not create from inner purpose, they do not truly participate in a cultural field, and they do not have a self to express.

For educators and researchers, here is what we need to keep in mind: AI can many great things including providing help with brainstorming, style adaptation, refining language, etc.

But we should be careful not to confuse fluent generation with full creativity. The real value of AI, therefore, is in helping us extend the conditions under which we can think, revise, explore, and create.

Reference:
Franceschelli, G., & Musolesi, M. (2025). On the creativity of large language models. AI & Society, 40, 3785–3795.

A new handbook worth adding to your AI and education reading list:

Handbook of Critical Studies of Artificial Intelligence and Education, edited by Wayne Holmes.

The volume has critical orientation which can serve best for critical AI literacy work. It does not simply ask how AI can make education faster, more efficient, or more personalized. It asks deeper questions:

Who benefits from AI in education?
Whose goals shape AI systems?
What happens to student agency, teacher labour, democratic learning, equity, and governance?
How do we move beyond hype and fear toward critical AI literacy?

The chapters cover a wide range of timely issues, including AI literacy, assessment, student voice, data colonialism, AI ethics, academic labour, policy, governance, and the sociocultural implications of AI in education.

The AI in education sphere is flooded with tools guides and adoption reports. Critical frameworks like these are needed. They help us teachers and researchers examine the assumptions, values, and power relations built into AI systems.

Reference:
Holmes, W. (Ed.). (2026). Handbook of critical studies of artificial intelligence and education. Edward Elgar Publishing.

When ChatGPT is gone, creativity reverts but its impact remains!

That is one of the main findings from a paper by Liu, et al. ( 2024). The researchers ran a seven-day lab experiment with university students, followed by a 30-day follow-up.

One group used ChatGPT for creative tasks during the middle five days. The other group completed the same tasks without ChatGPT.

While ChatGPT was available, students performed better. They generated more novel, useful, flexible, and polished ideas. In problem-solving tasks, their responses were also rated higher for creativity and quality.

So yes, ChatGPT gave them a boost. But here is the catch.

When ChatGPT was removed on Day 7, the creative advantage disappeared. One month later, the same pattern remained. The students who had used ChatGPT did not retain the creative gain when they worked independently.

Even more interesting, the homogeneity did not disappear.

The ChatGPT group’s ideas remained more similar, even after the tool was gone. In the authors’ words, participants continued to produce more AI-like responses without using ChatGPT.

That is the part that matters most for education.

AI can help students produce better ideas in the moment. But if students start with AI every time, they may not strengthen their own creative judgment. They may also begin to think inside the patterns AI repeatedly offers.

This is why I prefer a simple classroom sequence:

First, students generate their own ideas.

Then they use AI to challenge, expand, test, or refine those ideas.

Then they compare what changed.

The goal is to protect the student’s own creative starting point.

AI can support creativity, but students still need to build the muscles that remain when the tool is gone.

Link in the first comment!



References

Liu, Q., Zhou, Y., Huang, J., & Li, G. (2024). When ChatGPT is gone: Creativity reverts and homogeneity persists.

AI in education is not mainly a technology problem. It is an alignment problem as well.

A new BCG/Bett report argues that many education systems already have AI ambitions, but struggle to turn them into real action because ministries, schools, employers, funders, and technology partners are not always moving in the same direction.

The report highlights four things successful systems are doing:

1. Setting a clear ambition linked to workforce and economic needs
2. Investing in people, especially teachers and adult learners
3. Removing barriers around infrastructure, funding, procurement, and policy
4. Scaling beyond isolated pilots

One point stood out to me: AI adoption should not be treated as a tool rollout. It requires whole-system redesign: curriculum, assessment, teacher training, data infrastructure, accessibility, privacy, and governance all need to move together.

The future of education will not be shaped by AI alone. It will be shaped by how well institutions prepare people, build trust, and create systems that support responsible use.

Reference:
Bankert, L., Westrin, C., Goel, S., Mwangi, S., & Stepanenko, A. (2026). From ambition to action: Redesigning education for an AI-driven economy. Boston Consulting Group & Bett.

I revisited the classic paper The Weirdest People in the World? by Henrich, Heine, and Norenzayan, and I think its message is even more important in the age of AI.

The paper makes a powerful point: much of what we know about “human behavior” in psychology comes from a very narrow group of people: Western, Educated, Industrialized, Rich, and Democratic (WEIRD)societies. In many cases, the participants are American university students.

The problem is that these groups are not always representative of humanity. In fact, the paper shows that WEIRD people are often outliers in areas such as perception, fairness, cooperation, moral reasoning, self-concept, and analytic thinking.

This matters deeply for AI.

AI systems are trained on human data, but whose human data?

If the data overrepresents certain languages, cultures, values, communication styles, and ways of thinking, then AI systems may quietly reproduce a narrow view of humanity while presenting it as universal.

This is why AI literacy cannot be only about prompts, tools, and productivity. It also has to be about culture, representation, bias, and power.

When an AI system gives us an answer, we should ask:

Whose knowledge is centered?
Whose language is normalized?
Whose values are treated as default?
Whose ways of thinking are missing?

The lesson from this paper is simple: we should be careful when turning a narrow sample into a universal claim.

And in the age of AI, that warning becomes even more urgent.



Henrich, J., Heine, S. J., & Norenzayan, A. (2010). The weirdest people in the world? Behavioral and Brain Sciences, 33(2–3), 61–83.

The Government of Canada has recently released Canada’s National Artificial Intelligence Strategy: AI for All.

The strategy includes plans to create a National AI Literacy Initiative, reach 1 million entry-level post-secondary students, and train more than 3,000 educators with AI learning kits for classrooms.

The report also places special emphasis on trust, safety, privacy, inclusion, Indigenous leadership, Canadian culture, and responsible adoption.

Link in the first comment!