This week's episode dives into a wave of new research shaping how AI is actually being used in education. We explore what works (and what doesn't) when it comes to AI-generated feedback, including why blended, "hybrid" feedback may be the most effective approach - and why more feedback doesn't always lead to better outcomes. The conversation then turns to one of the most important emerging issues: bias in AI systems. From subtle differences in tone to stereotyping based on student characteristics, the research highlights why educators need to be cautious about the data they provide AI tools. "If you use AI to write feedback, it does not treat every student the same way equally." We also talk about the growing evidence around AI tutors - where they outperform humans, where they fall short, and what actually drives meaningful learning gains. Along the way, we tackle major questions around detection, student use, teacher workload, and whether AI can ever replace human connection. The big takeaway? AI is powerful. And how we design, guide, and use it in education matters more than ever.

Research Papers discussed this week

AI for Feedback

• Directive, metacognitive, or a blend of both? A comparison of AI-generated feedback types on student engagement, confidence, and outcomes https://doi.org/10.1016/j.caeai.2026.100553
• AI assistance in peer feedback provision: Pedagogically sound, but minimally adopted https://www.sciencedirect.com/science/article/pii/S0360131526000291
• Marked Pedagogies: Examining Linguistic Biases in Personalized Automated Writing Feedback https://arxiv.org/abs/2603.12471

AI and Bias

• The Life Cycle of Large Language Models: A Review of Biases in Education https://bera-journals.onlinelibrary.wiley.com/doi/10.1111/bjet.13505

AI Tutors

• AI tutoring can safely and effectively support students: An exploratory RCT in UK classrooms https://arxiv.org/abs/2512.23633v1
• LearnMate: Enhancing Online Education with LLM-Powered Personalized Learning Plans and Support https://dl.acm.org/doi/10.1145/3706599.3719857
• Effective Personalized AI Tutors via LLM-Guided Reinforcement Learning https://papers.ssrn.com/sol3/papers.cfm?abstract_id=6423358
• Unifying AI Tutor Evaluation: An Evaluation Taxonomy for Pedagogical Ability Assessment of LLM-Powered AI Tutors https://arxiv.org/abs/2412.09416v1

AI Detection

• Trusting AI to detect AI? A systematic evaluation of the reliability and robustness of current AIGC detection tools for student academic work (paywalled) https://www.sciencedirect.com/science/article/abs/pii/S0360131526000540

Teacher Workload

• Shiksha Copilot: Teacher-AI Collaboration for Curating and Customizing Lesson Plans in Low-Resource School https://arxiv.org/pdf/2507.00456v3

Student use

• The Secret Life of Students project - WonkHE Feb/March 2026 https://wonkhe.com/wp-content/wonkhe-uploads/2026/03/Wonkhe_SLOS2026_Jim_slides.pdf
• Is a random human peer better than a highly supportive chatbot in reducing loneliness over time? https://www.sciencedirect.com/science/article/pii/S0022103126000417?dgcid=rss_sd_all

Tina Austin joins Ray and Dan for a wide-ranging conversation about what AI adoption really looks like in US education beyond the hype, the headlines and the endless frameworks.

Tina is an educator, consultant, policy adviser and the founder of GAInable. She works with schools, colleges and faculty teams on responsible AI adoption. In this episode, she shares how her work evolved from teaching bioethics and AI ethics into supporting educators across the US as they grapple with policy, privacy, assessment, tools, and changing classroom practice. The conversation explores the fragmented reality of AI in education, why many teachers are feeling "frameworked out", and why Tina believes the best place to start is not with the tool, but with the problem you are trying to solve. We also dig into Tina's "UnBlooms" framework - a challenge to linear interpretations of Bloom's Taxonomy in the age of generative AI - and discuss critical thinking, student reflection, equity, privacy, and why educators should stay sceptical of easy answers. A thoughtful episode on using AI well, asking better questions, and meeting learners where they are.

Here's all the links you need:

Tina's new venture, Gainable AI Gainable.ai

The UnBlooms™ model: A Problem-Centered Framework for Learning Design in the AI Era https://zenodo.org/records/17298679

The UnBlooms™ Workbook: How to Design, Teach, and Assess Human Reasoning in the AI Era Tina's Custom GPT for UnBloom: Unbloom-it

You can contact Tina on email at [email protected]

In this AI News episode, Dan and Ray explore the fast-moving reality of AI in the workplace - and why many of us might not be as prepared as we think.

They unpack a striking story of a KPMG partner fined for using AI to cheat on an AI ethics course, raising questions about assessment, responsibility, and what "cheating" even means in an AI-enabled world.

The conversation then shifts to a growing trend: organisations and universities rolling out AI tools like Copilot at scale, and what this means for equity, productivity, and expectations in the workplace.

Dan and Ray also dive into new research from Anthropic on the future of work, highlighting which roles are most exposed to AI disruption - and why knowledge work may change more than hands-on professions. They explore the idea that the real shift isn't job loss, but job transformation.

Finally, they tackle bold claims from industry leaders that office roles could be automated within 12–18 months, debating what's hype versus reality.

This episode challenges a simple narrative: AI isn't just replacing work - it's redefining what valuable work looks like.

Links to discussions in the podcast:

KPMG partner fined for using artificial intelligence to cheat in AI training test https://www.theguardian.com/business/2026/feb/16/kpmg-partner-fined-artificial-intelligence-ai-training-test

Aston University is first Midlands University to provide all staff with Microsoft 365 Copilot and Copilot Chat for Students https://www.aston.ac.uk/latest-news/aston-university-first-midlands-university-provide-all-staff-microsoft-365-copilot-and

NY Times: Who's a Better Writer: A.I. or Humans? Take Our Quiz. https://www.nytimes.com/interactive/2026/03/09/business/ai-writing-quiz.html

Anthropic and US Government https://www.anthropic.com/news/statement-department-of-war

Anthropic future of labour report https://www.anthropic.com/research/labor-market-impacts

AI Exposure of the Australian & US Job Market AUS: https://0xtreme.github.io/aus-jobs/ US: https://joshkale.github.io/jobs/

Anthropic Education Report - AI Fluency index https://www.anthropic.com/research/AI-fluency-index

The end of office workers as we know it? https://www.msn.com/en-gb/money/other/microsoft-ai-ceo-mustafa-suleyman-issues-18-month-warning-for-office-workers/ar-AA1WvXMi?ocid=entnewsntp&pc=U531&cvid=69957afbd4c747f08d77afaead77499c&ei=56

Professor Stephen Heppell joins Dan and Ray for a wide-ranging conversation about the future of schools, assessment, and learning in the age of AI. Stephen reflects on more than four decades of innovation in education technology — from early experiments with AI and HyperCard through to today's generative AI systems. Drawing on work around the world, he shares stories from radical learning environments including beach schools, post-hurricane classrooms in the Cayman Islands, and experimental learning spaces designed with students themselves. A central theme of the episode is the growing gap between how schools currently operate and the skills the modern world demands. Stephen argues that as AI makes knowledge abundant, the most valuable human capabilities will be creativity, ingenuity, collaboration, and ethical judgement - qualities that traditional assessment systems rarely measure well. The discussion also explores how AI can support teachers rather than replace them, helping with differentiated learning activities, analysis of student understanding, and freeing teachers to focus on the human side of education. Finally, Stephen challenges educators and policymakers to rethink learning spaces, assessment, and student agency - and to build education systems that prepare learners for a rapidly changing world. If you want to read about more of Stephen's work, there's plenty more detail on Lindfield Learning Village and lots more on https://www.heppell.net/

In this news-packed episode, hosts Ray and Dan explore Purdue University's bold new requirement for all graduates to demonstrate AI competency; and the strategic partnerships between Harvey.ai (the specialised system for the legal profession) and universities in Sydney, Oxford and Chicago.

The conversation turns to the "first in the world" move by the University of Manchester to provide Microsoft 365 Copilot to 65,000 students and staff - paying homage to the legacy of Alan Turing.

A highlight of the episode is the deep dive into "vibe coding"— the phenomenon of non-programmers using AI to build applications through iterative prompting rather than manual syntax. They also tackle the "AI bubble," the rise of "work slop," and the surprising research showing that Boomers often have a more accurate understanding of how AI works than Millennials.

Links & Resources:

Purdue University adds 'AI working competency' graduation requirement https://www.purdue.edu/newsroom/2025/Q4/purdue-unveils-comprehensive-ai-strategy-trustees-approve-ai-working-competency-graduation-requirement/

University Law Schools introduce AI partnerships https://www.afr.com/companies/professional-services/sydney-and-uts-law-schools-bow-to-ai-wave-partner-with-harvey-20260119-p5nv49

University of Manchester announces 'world first' AI rollout with Microsoft https://www.manchester.ac.uk/about/news/world-first-ai-partnership-between-the-university-of-manchester-and-microsoft-announced/

"What we are doing about AI at UWA" https://www.uwa.edu.au/news/article/2026/february/what-we-are-doing-about-ai-at-uwa

High school students forced to fight false allegations of AI cheating https://www.abc.net.au/news/2025-12-22/ai-detectors-incorrectly-brand-high-school-students-ai-cheats/106138394

New Future of Work Report from Microsoft https://www.microsoft.com/en-us/research/wp-content/uploads/2025/12/New-Future-Of-Work-Report-2025.pdf

The Impact of AI on Work in Higher Education https://www.educause.edu/research/2026/the-impact-of-ai-on-work-in-higher-education

Americans Have Mixed Views of AI – and an Appetite for Regulation https://www.searchlightinstitute.org/research/americans-have-mixed-views-of-ai-and-an-appetite-for-regulation/

And finally....

From the "Do you ever read T&C's" dept https://www.linkedin.com/posts/matthewwemyss_i-logged-into-ai-studio-yesterday-and-i-ran-activity-7411400400177729536-hgPL

AI in Universities: Why Connection, Not Content, is Now King

This was an exciting episode, because we recorded it on campus at the world's newest university - Adelaide University. It officially started on-campus delivery this week, as it finally opened the doors after merging the University of Adelaide and the University of South Australia. Amid the buzz of students arriving for week 1, Eddie Major and I found some time to sit down and talk about how AI is impacting universities. Eddie is the university's AI Learning and Teaching Coordinator and you may not be surprised to learn that we discussed the AI myths of higher education and what being an "AI-first university" means.

Eddie debunks the "AI brain rot" myth, explaining that while the technology is disruptive, it is not the end of the university. Instead, we are moving from an era where "Content is King" to one where "Connection is King." We explore:

• Upstream AI Use: How students are using tools like NotebookLM to synthesise information before they even start an assignment.
• The Soft Skills Surge: Why communication and critical thinking are now more valuable than hard technical skills.
• The AI-First University: What it truly means to embed AI literacy across a global curriculum.

References:

We discussed the research from Hiromu Yakura at the Max Planck Institute about the way that ChatGPT was influencing speech. The paper, called "Empirical evidence of Large Language Model's influence on human spoken communication" is available at this link: https://arxiv.org/abs/2409.01754

• Can synthetic avatars replace lecturers? An exploratory international study of higher education stakeholder perceptions| https://link.springer.com/article/10.1186/s41239-025-00568-4
• Who grades best? Comparing ChatGPT, peer, and instructor evaluations across varying levels of student project quality https://www.tandfonline.com/doi/full/10.1080/02602938.2025.2588682?src=
• The Artificial Intelligence Disclosure Penalty: Humans Persistently Devalue AI-Generated Creative Writing https://doi.org/10.1037/xge0001889
• The older "Transparency Dilemma" paper referenced too: https://www.sciencedirect.com/science/article/pii/S0749597825000172
• Asking generative artificial intelligence the right questions improves writing performance https://www.sciencedirect.com/science/article/pii/S2666920X25000141?via%3Dihub
• When AI only asks: how question-driven dialogue shapes prewriting in the classroom https://www.frontiersin.org/journals/education/articles/10.3389/feduc.2026.1740044/full
• Prompt Repetition Improves Non-Reasoning LLMs https://arxiv.org/html/2512.14982v1
• How to Use Generative AI in Educational Research https://www.cambridge.org/core/elements/how-to-use-generative-ai-in-educational-research/916142E735B678F86A59240BFE651F5C
• PaperBanana: Automating Academic Illustration for AI Scientists https://dwzhu-pku.github.io/PaperBanana/ https://arxiv.org/abs/2601.23265

• OECD: Digital Education Outlook 2026
• UK DfE: Generative AI Product Safety Standards for Education (Jan 2026)
• UK DfE: Commitment to AI Tutoring for disadvantaged children
• NY Times: As Schools Embrace A.I. Tools, Sceptics Raise Concerns
• Deakin University's FutureFocus GenAI program
• Free E-Book: Teaching AI Ethics by Leon Furze (teachingaiethics.com)

In this special "flipped" episode, the tables are turned on your usual hosts, Dan Bowen and Ray Fleming. Interviewed by Dr. Michael Hallissy from (and for) the teachnet Ireland podcast, Dan and Ray step into the guest seats to share the "AI in Education" podcast origin story - from its 2019 "skunkworks" beginnings at Microsoft to its current status as an independent voice in the global edtech conversation.

The trio dives deep into how the podcast evolved through the 2022 Generative AI explosion, moving from technical "hoodie" discussions about algorithms to essential human skills like empathy and questioning. They reflect on impactful moments, including the complexities of indigenous data rights and why "AI detectors" are a failing tool for schools.

Beyond the backstory, Dan and Ray discuss the widening AI equity gap and their vision for 2026: a focus on "the doers"—the teachers implementing AI in the classroom today. Whether you're a long-time listener or new to the show, this episode offers a rare, personal look at the mission behind the mics.

We think this episode might be especially interesting to all the new listeners who have joined us over the last 6 months, who might have some questions about where and when the podcast started, and Ray and Dan's background! We want to especially thank Michael for asking us great questions, and Pat Brennan for being the technical & scheduling mastermind to make it happen.

Links & References TeachNet Ireland Podcast: https://teachnet.ie/category/podcasts/

• Apple: https://podcasts.apple.com/us/podcast/teachnet-podcasts/id1650615051
• Spotify: https://open.spotify.com/show/4hiz0yCcT7D5qs8J85Fl5K

Research Paper discussed: "Heads I Win, Tails You Lose"

Welcome to our first episode of 2026. In this heavy-hitting season opener, hosts Dan and Ray are joined by Dr. Mark Bassett, Academic Lead for AI at Charles Sturt University and a "superhero" of AI activism. Mark's an ally in our long standing mantra on the podcast, as we know you've got tired of hearing just Dan and Ray say "AI detectors don't work".

Dr. Bassett breaks down his landmark paper, "Heads We Win, Tails You Lose: AI Detectors in Education" which we describe (hopefully) as the final 'silver nail in the coffin' for detection software. We move past the surface-level "they don't work" argument and dive into the legal, ethical, and systemic risks universities face by relying on "black box" algorithms. Mark compares current AI detection to using a deck of tarot cards to determine a student's future - arguing that these tools have no place in a fair academic integrity process.

We also explore the S.E.C.U.R.E. framework, a tool-agnostic approach to integrating AI into education safely. If you're an educator, student, or leader wondering how to move from suspicion to capability-building, this is the blueprint you've been waiting for.

1. The Research Paper: Heads We Win, Tails You Lose: AI Detectors in Education

2. The Framework: The SECURE Framework for AI Integration

3. Find Mark Bassett online via his website and LinkedIn

4. Referenced Study: University of Reading's "Turing Test" paper on AI in Exams