Educators Technology

The teachers who get the most from AI are rarely the ones with the longest list of tools.

They are the ones who can explain why a particular tool belongs in a particular lesson, what learning goal it supports, and where it might interfere with the thinking students need to do themselves.

This idea is at the centre of my updated 68-page guide, The Educator’s Guide to AI Tools.

The guide brings together practical tools for:

Lesson planning
Science and mathematics
ESL and ELL instruction
English language arts
History and social studies
Art and music
Feedback and assessment
Rubric and worksheet creation
Teacher AI literacy and professional learning

Yes I included various AI tools but still I do not want teachers to treat this as another list they are expected to work through.

Choose one or two tools that address a real need in your classroom. Learn how they work. Try them with a clear learning goal. Pay attention to what happens.

One practice I recommend in the guide is writing a short reflective memo after using a new AI tool with students.

Did it produce what you expected?
How did students respond?
Did it help them think more deeply?
Did it remove a useful struggle?
How much editing did the output require?
What would you change the next time?

After several weeks, these memos begin to reveal patterns. You develop a personal evidence base grounded in your own classroom, your students, your subject, and your teaching style.

AI platforms will continue to change but your pedagogy gives you something more stable.

What do my students need to learn? How will they learn it? Where can AI support that process? Where might it get in the way?

Once these questions are clear, choosing a tool becomes much easier.

The guide is free to download. It is also licensed under Creative Commons, so educators can share and adapt it for noncommercial purposes with proper attribution.

Link in the first comment!



Reference

Kharbach, M. (2026). The educator’s guide to AI tools: Practical tools for lesson planning, feedback, assessment, and beyond. Educators Technology.

Here are three practical literacy activities to try with your students!

There is a simple way to bring ChatGPT into the literacy classroom without asking it to do students’ reading or writing: place its response in front of them and make it the object of investigation.

This practical article by Amy Hutchison offers three especially useful approaches.
Before reading a book, students can ask ChatGPT for an overview, identify unfamiliar words and ideas in its response, and ask follow-up questions.

The class can then discuss what they learned, what remains unclear, and which claims they want to examine while reading the actual book. In this role, ChatGPT helps students activate and build background knowledge.

Students can also analyze an AI-generated response as they would any other text. What information is missing? Which parts are vague? Does the answer contain bias?

How could it be improved?

One example in the article asks ChatGPT to explain how the illustrations in The Giving Tree contribute to the story. Students compare its response with the book and discover details the chatbot overlooked.

They have to return to the text, examine the illustrations carefully, and construct a stronger interpretation of their own.

The third activity teaches lateral reading. Students ask ChatGPT a factual question, leave the chat, consult several websites, and compare what those sources say.

They may discover factual errors, but they may also find that a seemingly accurate response omitted important information.

What I like about these activities is the role assigned to the student. The student remains the reader, investigator, evaluator, and writer. ChatGPT supplies material that gives them something to question.

Hutchison also encourages teachers to assess the writing process through drafts, revisions, and students’ explanations of how they used AI. This provides much richer evidence of learning than trying to guess whether a polished final paragraph came from a chatbot.

Young learners are still developing their evaluation skills, so teacher guidance is essential. They need us to model how to check a claim, recognize a vague answer, compare sources, and decide whether an AI suggestion deserves a place in their work.

AI becomes educationally useful when its output creates more reading, questioning, discussion, and judgment.

Reference:
Hutchison, A. (2024). Making artificial intelligence your friend, not your foe, in the literacy classroom. The Reading Teacher, 77(6), 899–908. https://doi.org/10.1002/trtr.2296

Brown University has released a 51-page report on generative AI in teaching and learning, and one tension runs through the entire document.

Students are using AI regularly. They are also deeply worried about what it may be doing to their learning.

The committee collected feedback from nearly 700 members of the Brown community, reviewed policies from peer institutions, examined course syllabi, and screened 148 scholarly and professional publications.

Among the students who responded, 56% of undergraduates and 67% of graduate and medical students reported using generative AI daily or weekly.

Many were using it to learn. They asked AI to explain difficult problems, assess their understanding, revise written work, summarize readings, and debug code.

Yet 88% of undergraduate respondents and 73% of graduate and medical student respondents expressed concern that AI could reduce their long-term learning. Similar numbers worried about its effects on their cognitive abilities.

Even among undergraduates who used AI daily, 78% were concerned about reduced learning.

I find this tension more useful than the familiar division between people who support AI and people who oppose it.

Students can recognize the value of a tool while remaining uneasy about how it changes their intellectual habits. They may appreciate an immediate explanation and still wonder whether they could explain the same concept later without assistance.

The report connects this problem directly to AI literacy.

Brown argues that AI literacy should include technical knowledge, critical evaluation, ethical reasoning, and an understanding of AI’s environmental, social, cultural, and political effects.

It also adds something I consider especially important: metacognitive awareness.

Students need to distinguish their own cognitive contribution from the contribution made by AI. They need to recognize which uses lead to durable learning and which allow them to avoid the intellectual work through which learning develops.

The report also raises concerns about the social experience of learning.

When students repeatedly ask a chatbot for explanations, they may begin replacing support from classmates and instructors. They lose conversations in which they question an idea, explain their reasoning, receive criticism, and work through uncertainty with another person.

Unclear or highly punitive AI policies can make this worse. According to the report, they may create secrecy, shame, and distrust. Brown recommends clear expectations accompanied by explanations. If AI is required, students should understand how it supports learning. If its use is limited, they should understand what knowledge or ability the restriction is protecting.

The committee also examined AI-detection tools.

Among the peer institutions it reviewed, none endorsed their use. The report points to false positives, false negatives, and bias against multilingual writers. These systems also cannot reliably distinguish between limited grammatical editing and having AI structure and draft an entire assignment.

Link in the first comment!



Reference

Brown University, Generative AI in Teaching and Learning Committee. (2026, July). Generative AI in teaching and learning (GAITL) committee final report and recommendations.

Sycophancy is when an AI tells you what you want to hear instead of what is true.

Anyone who has used a chatbot has felt this. You push back, the model folds. It apologizes, rewrites a correct answer, and agrees with you.

The problem for us as teachers is that sycophancy quietly teaches students that confidence equals accuracy. State a wrong answer firmly and the assistant will often repeat it back to you. Do that a hundred times across a semester and what students build is not knowledge but a habit of being agreed with.

In this paper, Sharma and colleagues tested five assistants: Claude 1.3, Claude 2, GPT-3.5, GPT-4, and LLaMA 2. A few findings worth highlighting here:

Ask "Are you sure?" after a correct answer and Claude 1.3 admitted a mistake 98% of the time. Accuracy dropped by up to 27% when a user suggested the wrong answer, even with hedging like "I'm really not sure."

Feedback shifted too. Say you wrote the essay and the comments got warmer. Say you disliked it and they turned critical. Same text, different praise.

Show the assistant a John Donne poem, tell it Sylvia Plath wrote it, and it will analyze Plath.

The cause is upstream. When the researchers analyzed human preference data, matching the user's beliefs was one of the strongest predictors of which response people chose. We trained this in.

Link in the first comment.

AI literacy is much more nuanced than what many teachers think!

In this paper, nnapureddy, Fornaroli, and Gatica-Perez provide an interesting framework containing twelve competencies that define AI literacy.

This is one of the most comprehensive AI literacy frameworks I have seen so far and guess what? Prompting is only one of them.

At the foundation is an understanding of AI itself: recognizing different forms of AI, knowing at a basic level how generative models work, and understanding that a chatbot generates content through statistical patterns. It does not retrieve and verify information in the same way as a search engine.

AI-literate users also need realistic knowledge of what these tools can and cannot do. They should understand hallucinations, bias, privacy risks, deepfakes, impersonation, and the possibility that an answer may combine accurate information with complete fabrication.

The practical competencies include choosing and using appropriate tools, critically assessing their outputs, writing effective prompts, and recognizing AI-generated content while understanding the serious limitations of AI detectors.

The framework also includes programming and fine-tuning models. The authors make an important clarification here: most users will never need this advanced technical competency to become AI literate.

Surrounding these technical skills are questions of context, ethics, and law. Is AI appropriate for this particular task? What are the institution’s rules? Who owns the generated material? Was private information entered into the system? Could the output mislead, discriminate, or cause harm?

The twelfth competency is the one that holds everything together: the ability to learn continuously. Tools change, models improve, regulations develop, and new risks appear. AI literacy has no final point of completion.

Learn more about this framework from the link in the first comment!

Reference:
Annapureddy, R., Fornaroli, A., & Gatica-Perez, D. (2025). Generative AI literacy: Twelve defining competencies. Digital Government: Research and Practice, 6(1), Article 13. https://doi.org/10.1145/3685680

Plagiarism used to leave a visible trail.

A student copied words or ideas from an identifiable source and presented them without acknowledgment.
Generative AI has made that trail much harder to follow because its involvement can occur throughout the entire process.

A student might use AI to brainstorm, reorganize an argument, translate a passage, improve a sentence, generate examples, or write the whole assignment. Each case involves AI, but they do not carry the same meaning for learning or academic integrity.

Cecilia Chan (2023) uses the term AI-giarism, which she defines as “an emergent form of academic dishonesty involving AI and plagiarism” (p. 1).

The term gives us a starting point. The difficult work begins when we try to decide what actually counts as misconduct.

I use AI myself to explore ideas and express them more clearly. Many students and academics do the same. The presence of AI does not tell us who developed the argument, evaluated the evidence, made the decisions, or performed the intellectual work.

This is why I find it more useful to think of AI use as a continuum.

At one end, AI assists with a limited part of the work while the person remains intellectually responsible. As its role expands, assistance can become collaboration, collaboration can become delegation, and delegation can eventually become proxy performance: the tool appears to demonstrate knowledge or ability that the student has not developed.

The boundary will change across tasks and disciplines. Using AI to improve grammar in a written reflection is different from asking it to generate the reflection. Using it to question an argument is different from allowing it to construct the argument from beginning to end.

The work of Eaton on postplagiarism, Corbin and colleagues on acceptable AI use, Dawson and colleagues on assessment validity, and Nieminen and Eaton on accommodations all points toward the same unresolved problem.

Our inherited definitions of plagiarism cannot carry this discussion alone. We also need to examine authorship, agency, disclosure, assessment validity, and who performed the intellectual work being assessed.

Academic integrity has become harder to define. That is precisely why we need clearer language, better assessment design, and more honest conversations with students.

Link in the first comment!



Reference

Chan, C. K. Y. (2023). Is AI changing the rules of academic misconduct? An in-depth look at students’ perceptions of “AI-giarism.” arXiv.

AI assistance can make students faster. Does it also make them better thinkers?

This question is at the centre of a new study by Yuanyan Hu et al.

The researchers followed 102 Chinese undergraduates through three AI-assisted writing tasks in a 16-week English for Academic Purposes course.

Students compared their translations of research abstracts with AI-generated versions. They used AI while revising a research introduction and reviewing the literature. They also interpreted a diagram, evaluated an existing data commentary, and rewrote it.

The students reported improvement in several areas of critical thinking, including identifying information, comparing ideas, questioning assumptions, forming hypotheses, and evaluating arguments. They also became more sensitive to relevance and logical coherence.

However, the gains were much smaller in areas such as depth, flexibility, precision, and the ability to monitor one’s own thinking.

That last one is especially important.

The researchers call it meta-critical thinking: the ability to step back, examine how you are reasoning, regulate the process, and judge the quality of your own thinking.

It was almost entirely absent from the students’ interactions with AI.

Most prompts focused on polishing language, translating text, organizing structure, or saving time. Very few asked AI to support logical reasoning or reflection.

Peer interaction produced a different pattern.

Students turned to one another to unpack difficult tasks, question causal connections, justify interpretations, evaluate arguments, and negotiate meaning. AI helped with operational work. The deeper reasoning often happened between people.

The type of task also made a difference.

The most cognitively demanding activity asked students to critique and reconstruct a research introduction. It required them to bring together sources, assess coherence, identify gaps, and build an argument. This task generated the strongest evidence of critical thinking.

A more procedural task involving data commentary made students efficient, but AI sometimes completed parts of the cognitive work for them.

The authors describe their proposed approach as “AI-triggered, peer-constructed.”

AI can provide an output, surface a problem, introduce another perspective, or give students something to question. Peers then discuss, challenge, justify, and construct the deeper understanding together.

For me, this is the most useful message from the study.

The presence of AI does not create critical thinking. The task has to demand it. Students need something worth questioning, enough openness to make decisions, and opportunities to test their reasoning with other people.

The study relied partly on student self-reports and task observations, so it cannot establish objective gains in critical-thinking ability. Still, it gives us a more useful question to bring into our classrooms:

What kind of thinking does this AI-assisted task actually require from the student?

Link in the first comment!



Reference

Hu, Y., Curdt-Christiansen, X. L., & Wang, J. (2025). Smart tools, smarter minds? Learner-AI interaction and AI assistance on critical thinking in EAP contexts. Journal of English for Academic Purposes, 78, Article 101586.

Perhaps the most important questions about AI in education are not about what the technology can do.

They are about what it changes.

What happens to education when AI systems begin shaping how knowledge is produced, how students are assessed, how teachers work, and how institutions make decisions? Whose interests do these systems serve? Which voices and values do they privilege and who might be excluded?

These are the questions at the heart of an emerging field called Critical Studies of AI and Education (CSAI&EDA).

In this important paper, Holmes and colleagues move the conversation beyond the familiar focus on adopting AI tools, improving productivity, or personalizing learning. They invite us to examine AI as a political, cultural, and institutional force, one that can reshape teaching, learning, educational institutions, and even democratic life.

The paper maps the central themes and concerns defining this developing field while also discussing the barriers facing scholars undertaking critical work on AI and education. It concludes with a strong call for interdisciplinary collaboration and collective action.

This is foundational reading for anyone who wants to understand not only how AI is being used in education, but what kind of education its use may help create.

Reference:

Holmes, W., et al. (2025). Critical studies of artificial intelligence and education: Putting a stake in the ground.

MIT has just released a 40-page report on AI in teaching and learning, and its main concern is much larger than cheating.

Students at MIT are already using AI widely. According to the report, some see it as a source of help, efficiency, and creative inspiration. Others feel anxious, pressured, and uncertain about what is allowed.

What caught my attention is how AI appears to be changing the social experience of learning.

The committee heard reports of fewer students attending office hours, participating in discussions, or meeting in study groups. When students turn immediately to a chatbot, they may miss the conversations through which they learn to explain an idea, disagree with someone, accept criticism, and work through confusion.

The report also takes assessment seriously. If AI can produce an essay, solve a problem, write code, or construct a proof, the finished product can no longer tell us enough about what a student understands.

MIT recommends oral exams, portfolios, project demonstrations, conversations about submitted work, drafts, version histories, and regular checkpoints. These approaches give teachers more opportunities to see how the student’s thinking developed.

One concept that appears throughout the report is productive struggle.

Students need to experience some difficulty. They need time to get stuck, try an approach, make a mistake, and revise their thinking. The report warns about “cognitive surrender”: the habit of turning to AI at the first sign of difficulty and allowing it to take over the thinking.

AI can still have an important place in learning. It can provide feedback, help students practise, make learning more accessible, and allow them to undertake projects that were previously too complex. The report describes this as augmentation. The student remains intellectually involved and responsible for the work.

MIT also calls for clear course-level AI policies. Students should know when AI is allowed, limited, required, or prohibitedand why. A single university-wide rule will not work equally well for a poetry seminar, a mathematics course, an architecture studio, and a software-engineering project.

The report advises caution with AI detectors because of false accusations and the distrust they can create. It also asks instructors to disclose when they use AI to create teaching materials, provide feedback, or evaluate student work.

I think this may be the most important part of the report: education is also a social and human practice.

Students learn through relationships, shared work, mentorship, discussion, disagreement, and the slow development of confidence and judgement. A chatbot may provide an answer, but it cannot give students the full experience of becoming members of an intellectual community.

AI-aware education will require much more than adding a paragraph to the syllabus. We need to reconsider what students should learn, how they should learn it, and what evidence will genuinely show that learning has taken place.

Link in the first comment!



Reference

MIT Ad Hoc Committee on AI Use in Teaching, Learning, and Research Training. (2026, August 13). Report of MIT’s Ad Hoc Committee on AI use in teaching, learning, and research training. Massachusetts Institute of Technology.

The final product tells us less than it used to.

I shared this paper some time ago, but I keep returning to its central idea: assessment needs to make student thinking visible.

For years, we placed enormous weight on what students submitted at the end: the essay, report, presentation, solution, or completed project. We assumed that the quality of the product gave us reliable evidence of learning.

Generative AI has weakened that assumption.

A student can now submit polished work while remaining largely absent from the thinking that produced it. This leaves me increasingly interested in the trail behind the final submission.

How did the student approach the task?

What prompts did they use?

Which AI suggestions did they accept, reject, or revise?

How did they check the accuracy of the information?

What decisions did they make?

What can they explain without returning to the chatbot?

Reflective writing can help us see some of this process. Students might use AI for brainstorming, drafting, or feedback, while also documenting what they did, critiquing the output, and explaining how their own understanding shaped the final work.

Authentic assessment is also becoming more important. Local problems, professional scenarios, clinical dilemmas, multimedia projects, and context-rich cases require students to interpret information and make decisions.

They also create opportunities for follow-up questions, demonstrations, and conversations about the work.

I have always found the phrase “AI-resistant assessment” interesting. Complete AI-proofing is no longer realistic. Students can access AI in too many places and use it in too many ways.

The goal is to design assessments that keep the learner intellectually present.

A strong final product still matters. We simply need more evidence showing who did the thinking, how that thinking developed, and what the student can now understand, explain, and do.

Link in the first comment!