What happens when AI helps students earn more A grades, and also contributes to more failures when it's taken away?

In this episode, Ray and Dan explore new research on AI's impact on student performance, assessment integrity, and learning. They discuss studies linking AI to rising grades, the risks of over-reliance on AI, and growing evidence that AI tutors may support learning better than general-purpose chatbots.

The conversation also covers AI detectors, AI humanisers, teacher workload, Microsoft's latest Copilot updates, and a new tool for measuring the environmental footprint of AI use.

• Artificial Intelligence and Grade Inflation https://escholarship.org/uc/item/80x8d3qd ps WSJ wrote an article on this https://www.wsj.com/us-news/education/a-grades-are-suddenly-everywhere-since-the-arrival-of-chatgpt-845baae7
• Failing grades soar as professors see greater AI usage, dwindling math skills in UC Berkeley computer science classes https://www.dailycal.org/news/campus/academics/failing-grades-soar-as-professors-see-greater-ai-usage-dwindling-math-skills-in-uc-berkeley/article_16fad0bf-02cb-4b8c-8d88-888ffd9f8608.html
• Building AI Companions that Prioritise Learning over Performance https://arxiv.org/abs/2605.04816
• Effective Personalized AI Tutors via LLM-Guided Reinforcement Learning https://papers.ssrn.com/sol3/papers.cfm?abstract_id=6423358
• Law Professors prefer AI over peer answers https://law.stanford.edu/wp-content/uploads/2026/06/salinas_et_al.pdf
• Fixing teachers' problems? exploring teachers' repair and maintenance work around generative AI technologies https://www.tandfonline.com/doi/full/10.1080/01596306.2026.2657793\
• Dramaturgies of Deception: AI Humanizers and the Performance of Legitimacy in Higher Education Assessment https://arxiv.org/abs/2605.02649
• Homogenizing effect of large language models (LLMs) on creative diversity: An empirical comparison of human and ChatGPT writing https://www.sciencedirect.com/science/article/pii/S294988212500091X
• AI Detectors Fail Diverse Student Populations: A Mathematical Framing of Structural Detection Limits https://arxiv.org/abs/2603.20254

• Copilot Cowork released https://www.microsoft.com/en-us/microsoft-365/blog/2026/06/16/copilot-cowork-is-now-generally-available/
• Copilot notebooks released for all free copilot chat basic accounts https://techcommunity.microsoft.com/blog/educationblog/copilot-notebooks-and-study-guide-now-available-to-copilot-chat-users/4527320
• Andy Masley's Carbon footprint calculator| https://www.andymasley.com/visuals/ai-prompt-footprint/

• Mazenod College: Year 12 students caught using AI to cheat https://www.abc.net.au/news/2026-06-09/year-12-students-in-melbourne-caught-cheating-using-ai/106777700

What does it mean to put humans first in the age of AI?

Recorded at the Flourish Conference, this special episode brings together three school leaders who are using AI in very different ways, and with a common focus on human agency.

You'll hear how students are designing their own learning agents, how AI can help teachers make better use of student data while keeping information secure, and how one school is using AI-powered coaching tools to support professional learning and culture change.

Across all three stories, a clear message emerges: the most effective AI initiatives aren't replacing people - they're helping students and teachers flourish.

Featuring:

Alison Perosin (Our Lady of Perpetual Succour, Broken Bay Diocese) Hamish Jefferies (John Therry Catholic College) Renée Williams (The Cathedral College, Rockhampton)

A practical look at how schools are building capability, trust and innovation with AI.

Why would a leading law school ban AI entirely while other countries are giving every citizen access to ChatGPT?

In this news-focused episode, Ray and Dan unpack some of the biggest developments shaping AI and education around the world. They discuss China's new national AI education strategy, Malta's ambitious "AI for All" programme, Harvard's expansion of student AI access, and Anthropic's $200 million partnership with the Gates Foundation.

The conversation explores a controversial decision by the University of California, Berkeley School of Law to prohibit AI use in assessed work, raising important questions about judgement, employability, and the future role of AI in professional education.

They also examine new research on how people are actually using AI, why Australian students' digital literacy is falling despite increased screen time, and what educators can learn from a high-profile academic integrity case involving an AI-assisted newspaper article.

Finally, they highlight Jason La Greca's excellent framework for testing and stress-testing educational chatbots before they are deployed to students.

All the links:

What is the purpose of education in an AI-driven world?

In this thought-provoking episode of the AI in Education Podcast, Ray and Dan sit down with Pasi Sahlberg to explore one of the biggest questions facing schools, universities, and society today.

Drawing on global research, OECD trends, and decades of educational leadership, Pasi explains why traditional measures of success - achievement, credentials, and test performance - may no longer be enough in the age of AI.

The conversation explores:

• Human capital vs human flourishing
• Why wellbeing and agency matter more than ever
• The future of assessment and PISA
• AI's impact on work, learning, and society
• Why "hope is not a strategy"
• What schools should prioritise over the next decade

The episode also reflects on parenting, teacher accountability, screen time, and the human skills that may become most valuable as AI capabilities accelerate.

Referenced in this episode:

OECD Education for Human Flourishing. https://www.oecd.org/en/publications/education-for-human-flourishing_73d7cb96-en.html

An accidental guru: The making of an education warrior https://www.sciencedirect.com/science/article/pii/S073805932600060X

Sir Ken Robinson at TED - Do Schools Kill Creativity? https://www.ted.com/talks/sir_ken_robinson_do_schools_kill_creativity

.

This special live episode of the AI in Education Podcast was recorded at the CEnet Future:Forward Conference "Flourish 2026", where Dan and Ray explored one of the biggest questions facing education today: how do schools find the "happy middle" with AI?

The conversation dives into the shifting narrative around AI and jobs, the growing role of human agency in education, and why wellbeing, flourishing and trust must remain central as AI adoption accelerates. Along the way, they unpack new research on AI bias, AI detectors, cognitive debt, student safety, and the widening gap between individual innovation and organisational readiness.

The episode also reflects on keynote insights from Pasi Sahlberg and discussions around OECD flourishing metrics, parent engagement, and what schools can do now to bring entire communities along on the AI journey.

This is a thoughtful, practical and deeply human conversation about balancing opportunity, risk and responsibility in education's AI future.

Topics covered:

• AI and the future of work
• Human flourishing and wellbeing
• AI bias in education
• Safe AI use in schools
• Parent and community engagement
• AI detectors and academic integrity
• The "happy middle" approach to technology adoption

Research Papers, and links to things we discussed

• The changing tune of the AI leaders: The Jobs Apocalypse no more...See these tweets for last year's story: Jensen Huang, Sam Altman, and Mustafa Suleyman And this year's story: Sam Altman, Mark Zuckerberg and Jensen Huang

• Microsoft's Work Trend Index report 2026 https://www.microsoft.com/en-us/worklab/work-trend-index/agents-human-agency-and-the-opportunity-for-every-organization
• Pasi Sahlberg His website: https://pasisahlberg.com/ OECD research he discussed: https://www.oecd.org/en/data/dashboards/pisa-education-and-skills/digital-leisure-outside-school.html (the chart was from Figure 2.4 here)
• Victoria Hedlund, the "AI Bias Girl' https://www.linkedin.com/in/victoriamhedlund/ and on Substack at https://victoriahedlund.substack.com/ Her LinkedIn post that kicked off the SquashMallow test: https://www.linkedin.com/posts/victoriamhedlund_biasgirl-biasaware-stem-activity-7454786540133584896-E64
• The retracted Nature research paper on AI in Education: https://www.nature.com/articles/s41599-025-04787-y
• Think U Know: https://www.thinkuknow.org.au/

Inside Sydney's AI Hub: Building University Automation That Works In this episode, Dan and Ray are joined by Dan Hart, now leading the Automation and Innovation Hub at the University of Sydney. Dan shares how the Hub has evolved from a robotic process automation team into a university-wide service using automation and AI to improve everyday work. With around 400 automated processes running across the university, the team helps staff remove repetitive, time-consuming tasks without taking away the human parts of their roles. The conversation explores practical examples, including AI-powered invoice processing, and how generative AI is changing the software development lifecycle. Dan explains how tools like Cursor and AI-assisted coding are speeding up development, while also raising important questions about security, code quality, workload intensity, and developer wellbeing. The episode also dives into vibe coding, what it means for enterprise systems, and why the future of software development may depend less on hand-writing code and more on communication, problem-solving, and understanding users.

We kick off Series 17 with a multi-interview episode!

Recorded live at the The Anglican Schools Corporation "Day of AI" in Western Sydney, this episode brings together three educators who are reshaping teaching, assessment and student learning in real classrooms with AI.

You'll hear from Maria Mertzanakis at Oran Park Anglican College on how teachers are building shared "Brains" in NotebookLM to save time, improve differentiation and support pedagogy, while reducing workload by more than 50 hours a week across staff workflows.

Nathan Jones from Marsden Park Anglican College shares how AI is opening new creative possibilities for students and teachers alike, and why schools needed to guide students toward using AI well rather than trying to ban it.

And Patrick Ell from Roseville College explores what happens in a "post-product world", where assessment can no longer rely on trust in the final submission alone. His focus? Designing learning experiences that value process, thinking and cognitive craft.

We had a lovely grounded, practical and optimistic conversation with these teachers, who are leading from the front as schools navigate the realities of AI in education. Really great to have been invited to attend and participate by Julian Ridden, the TASC Head of AI.

This week's episode explores a defining moment for education in the age of AI.

Fresh from a week of major events in Sydney, including the Microsoft AI Summit, Dan and Ray unpack the newly released Castlereagh Statement - a collaboration between 70+ education leaders calling for urgent change across schools, universities, and training systems. Their message is clear: education isn't ready for the speed and scale of AI disruption.

But that's only part of the story.

We're also seeing a growing pushback on technology in classrooms, with schools limiting screen time and universities questioning device use; while, at the same time, new AI-first models like AI-led tutoring and mastery-based learning are emerging rapidly.

Add to that a widening gap between what employers expect and what graduates can actually do with AI, plus billions being invested in infrastructure and skills, and it's clear: something has to change.

We also dive into the latest research on AI in teaching, from new peer learning models to rethinking assessment and feedback.

This episode connects the dots across policy, practice, and innovation.

Resources discussed this week

• Castlereagh Statement https://castlereagh.ai/
• California schools introduce tech limits https://www.theguardian.com/us-news/2026/apr/22/los-angeles-school-district-screen-time
• Yale considering banning all electronic devices https://www.timeshighereducation.com/news/device-free-classrooms-mooted-yale-plan-rebuild-public-trust
• Khan TED Institute announced https://blog.khanacademy.org/introducing-the-khan-ted-institute-a-new-approach-to-higher-education/ https://khanted.org/Home
• LSI: The world's first AI-led university https://lsi.ac.uk/
• Pearson/AWS survey of employers re students https://www.pearson.com/en-us/power-of-learning/ai-readiness.html
• The Pedagogical Promptbook https://edtechbooks.org/promptbook/
• Microsoft's $25 billion AI in Australia announcement https://news.microsoft.com/source/asia/features/investing-in-australias-ai-future/
• Harvard adds mandatory AI courses to English writing course https://www.thecrimson.com/article/2026/4/23/expository-writing-ai/

Research Papers

• Think–Pair–Chatbot–Share: AI-Facilitated Peer Learning in Chemistry https://pubs.acs.org/doi/full/10.1021/acs.jchemed.5c00438
• Identifying what our students have learned: a framework for practical assessment validation https://doi.org/10.1080/02602938.2026.2620053
• Using generative artificial intelligence to reimagine feedback in higher education: a collaborative autoethnography https://www.tandfonline.com/doi/full/10.1080/02602938.2026.2653889#abstract

In this episode, we're joined by Dawn Knight, a sustainability lead and specialist support educator working with deaf students in a UK secondary school (Epsom and Ewell High School). Dawn shares how she's using AI in simple, practical ways to make learning more accessible - not just for students with additional needs, but for everyone in the classroom. From generating transcripts and differentiated resources to using AI as a planning "checklist," Dawn explains how these tools are helping her save time and reinvest it where it matters most: supporting students directly. The conversation explores how technology originally designed for accessibility - like captions and transcription - is now benefiting all learners, and why inclusive approaches to AI can unlock better outcomes across the board. We also dive into the realities of AI adoption in schools: what works, what doesn't, and why starting small is often the best approach. Dawn offers a refreshingly grounded perspective, showing that you don't need to be a technical expert to make a real impact. If you're looking for practical ways to use AI to support students and improve classroom experiences, this episode is full of ideas you can start using today.

This might also be the ideal episode to share with any colleagues who are at the beginning of their AI in teaching journey

In this episode, Dan and Ray explore how AI is rapidly reshaping education, workplaces, and policy - often faster than institutions can respond.

They unpack new guidance from NSW's NESA on student AI use, highlighting the growing tension between rules, real-world behaviour, and the need for clearer, more consistent policies across education systems.

The conversation expands globally with insights from the 2026 Stanford AI Index Report, revealing that while over 80% of students are already using AI, formal education and policy frameworks are struggling to keep pace. The distinction between 'AI in education', 'AI literacy', and 'AI education' becomes critical for understanding what schools and universities could actually be building.

They also discuss emerging tools like Adobe Student Spaces, evolving AI workflows using tools like Claude and Copilot, and new data showing Australians are among the most advanced AI users globally.

Finally, they revisit "real AI" use cases - specifically wildlife (crocs, turtles and quokkas) detection in Australia - as a reminder that AI's impact extends far beyond chatbots.

News items discussed

• Scientists use drones, cloud, and AI to protect Australia's Quokkas https://news.microsoft.com/apac/features/preserving-diversity-with-ai/
• Indigenous knowledge and AI help protect baby turtles from predators on Australia's remote Cape York https://news.microsoft.com/apac/features/indigenous-knowledge-and-ai-help-protect-baby-turtles-from-predators-on-australias-remote-cape-york/