Episodes

  • AI Is Cheaper In English | Check-In 79
    Nov 17 2026

    In this ChatEDU Check-In: AI Is Cheaper in English :(, Matt explores new research showing that generative AI models consume significantly more energy and produce less accurate results when processing non-English text. Researchers evaluated three open-source models across 122 languages and discovered that roughly 5.2 billion people require more than double the energy of English speakers to generate text in their native languages. This disparity is driven by tokenization, as non-Latin scripts often break down into far more tokens, increasing both computational costs and user pricing.


    Key Takeaways:

    Asking an AI model a question in a non-English language takes anywhere from 1.4 to 180 times more energy, with languages like Hindi and Bengali requiring 7.9 and 15.3 times more energy than English.


    Non-English scripts, particularly those with combining characters and accents, break into more tokens, which demands more memory, reduces batch efficiency, and drives up token-based costs for users.


    Higher energy consumption correlates with lower output accuracy, creating a "double penalty" that risks pricing non-English speakers out of AI access unless prompts are first translated into English or routed to language-specific models.


    Matt’s Two Cents: This episode highlights a crucial intersection between technical AI literacy and global educational equity. Understanding how language models tokenize different scripts helps students and educators grasp the hidden computational and financial costs of AI. For district leaders serving multilingual communities, these findings underscore the need to advocate for equitable, culturally responsive AI tools that do not penalize non-English speakers with higher costs or degraded performance.


    Article:

    Generative AI uses more energy on non-English text

    https://tinyurl.com/5n7wpvwv


    Show More Show Less
    4 mins
  • Haidt and Horvath on Screens in Schools: What They Get Right and Wrong | Ep. 136
    Nov 13 2026
    You cannot raise a kid in 2026 on 1926 parameters and expect it to work. Jonathan Costa on the screen-time backlash in schools.First UpOpenAI would like folks to call data centers compute campuses. The rebrand landed the same week voters rejected data center proposals in 40 of 41 local ballot measures, which makes it a ready civics lesson for any classroom covering AI and energy.OpenAI Wants Everyone to Call AI Data Centers 'Compute Campuses': https://tinyurl.com/4tkuwr86 The RundownTextbooks in Gemini NotebookGoogle Just Gave Students Access to Hundreds of Textbooks in Gemini Notebook: https://tinyurl.com/yauvfpyp Google, Search and SchoolsChatGPT's Search Volume Is Now Roughly 20 Percent of Google's: https://tinyurl.com/58cbvk79 Google Will Turn Off Search for Schools That Reject AI: https://tinyurl.com/2cfdumty Tell Google: Don't Make Educators Decide Between Safety and Education: https://tinyurl.com/4syejrd2 Brisk's AI search blocker (YouTube): https://tinyurl.com/3yuy3xhy New York's Kids' AI LawGovernor Hochul Signs Strongest Kids' AI Safety Law in the Nation: https://tinyurl.com/msd5fyu2 Chromebook LawsuitParents Sue Google Over AI Safety Risks on School Chromebooks: https://tinyurl.com/bdcucw67 Tech SovereigntyCould Tech Sovereignty Free Higher Ed From Big Tech?: https://tinyurl.com/c5e2jtyc Journals Hit PauseSix Probability and Statistics Journals Pause Submissions to Rethink AI Policy: https://tinyurl.com/mr3jxyyx How to Deslopify Your AI: https://tinyurl.com/bddj3erv Phones and ScreensNEASC Leadership Pulse #3: Cell Phone Policies (LinkedIn): https://tinyurl.com/yhp6pb78 LBUSD Is Evaluating AI Use in the Classroom: https://tinyurl.com/4wz68bje Paolina Cummings at the Long Beach school board (Instagram): https://tinyurl.com/mvdk27pm Eastern Illinois University Student Senate Resolution 26-27-01 on AI in the classroom (LinkedIn): https://tinyurl.com/32ewej6d Beneath the SurfaceJonathan Costa, executive director of EdAdvance, on the screen-time backlash in schools. He grants that screens for their own sake are indefensible, then takes apart the claim that students are getting cognitively worse, pointing to PISA scores in Singapore and Estonia and to Peter Gray's Restoring Childhood as a rival reading of the same data.Restoring Childhood by Peter Gray (book): https://tinyurl.com/2znavwev Bright ByteGoogle put a small data center in orbit and asked the onboard Gemma 4 model for a haiku about space. It is a concrete way into a question students can actually chew on, which is where the energy behind AI comes from and what it costs.Google's First Data Center Satellite Is Online and Writing Poetry: https://tinyurl.com/yy6h8zhk AnnouncementsRegister for our Winter Microcredential and upcoming AI events: https://tinyurl.com/43jwkwypPurchase Learning They'll Love - Dr. Elizabeth Radday on Amazon: https://tinyurl.com/22t9hz77Check out our middle / high school Student AI Literacy course: www.skills21.org/ai/learnaiExplore Skills21’s FREE social media literacy curriculum: https://www.skills21.org/socialAddress Screen Time concerns: skills21.org/ai/screenshiftRead From Panic to Proof: A Data-Driven Approach to Screen Time (ASCD Educational Leadership): https://tinyurl.com/4pwdu68SponsorThe National Center for Next Generation Manufacturing: Bridging Education and Industry to Build Tomorrow’s Technical Workforce - https://www.nextgenmfg.orgRegistration Link: https://forms.gle/LeVh9FubGVY6BNuJ6. To learn more about the NCNGM, visit www.nextgenmfg.org
    Show More Show Less
    1 hr and 16 mins
  • Finland Halts Google Data Center | Check-In 78
    Nov 12 2026

    In this ChatEDU Check-In: Finland Halts Google Data Center, Liz explores criminal investigations into Google after the company cleared protected forest land for a new data facility. Regulators ordered early development halted across two properties due to bypassed environmental risk assessments and water management concerns. This episode highlights the mounting tension between aggressive artificial intelligence infrastructure expansion and regional environmental protection.


    Key Takeaways:

    Finnish authorities initiated a criminal inquiry into whether Google bypassed mandatory environmental reviews before beginning land clearance for a data facility.


    The clearing of hundreds of thousands of trees mobilized civic protests and petition drives, eroding public goodwill toward foreign technology investments.


    Expanding data infrastructure has intensified debate over whether high energy and land consumption offers meaningful domestic payoff to local communities.


    Liz’s Two Cents: As school districts rapidly expand their adoption of cloud-based digital platforms and artificial intelligence tools, leadership teams must look beyond vendor capability to consider the real-world ecological costs of data infrastructure. District leaders should use this moment to evaluate enterprise vendor sustainability commitments and help students understand the physical footprint behind the digital tools they use every day.


    Article:

    Finland launches criminal investigation into Google’s clearing of forest for datacentre site

    https://tinyurl.com/54za3hvj


    Show More Show Less
    2 mins
  • CDs Are So Back! | Check-In 77
    Nov 10 2026

    In this ChatEDU Check-In: CDs Are So Back!, Liz explores why music fans are leaving streaming platforms to buy CDs. As algorithmic curation and digital fatigue grow, listeners are increasingly turning toward physical media for deliberate and meaningful engagement. This episode highlights how the resurgence of compact discs reflects a wider desire to escape automated feeds and reclaim intentional ownership.


    Key Takeaways:

    Listeners are turning to physical albums to escape the endless decision paralysis and automated curation of digital platforms.


    Direct media purchases provide vital financial backing to creators and safeguard collections against disappearing digital licenses.


    Physical compact discs give listeners permanent ownership and a durable, affordable alternative to the high cost and delicate footprint of vinyl.


    Liz’s Two Cents: The growing pushback against algorithm-driven streaming offers a compelling lesson for school district leaders navigating digital learning environments. While infinite digital access provides undeniable convenience, it frequently creates cognitive overload and passive scrolling habits among students. School systems must strike a thoughtful balance by pairing digital literacy with intentional, focused offline experiences, helping students move past digital fatigue to build deep, sustained engagement.


    Article:

    ‘Algorithms make everything predictable’: why the CD is making a comeback

    https://tinyurl.com/864u84wf


    Show More Show Less
    3 mins
  • How 250,000 Teachers are Fighting AI Cheating with Process, not Punishment | Ep. 135
    Nov 6 2026
    ChatGPT can write a solid essay in seconds, so the finished draft doesn’t prove much anymore. The founder of Revision History and two amazing educators argue that what counts now is watching the work happen and making thinking visible. First UpMatt tries to gift Liz a $499 Dyson CameraJet, a Wi-Fi toothbrush whose camera reads 28 images a second, finds the gaps and blasts them with mouthrinse. You can livestream it. She declines.Dyson puts AI in a $499 toothbrush: https://tinyurl.com/m7sm4hdm The RundownCanva and Common SenseCanva adds 60 free AI and digital literacy lessons: https://tinyurl.com/4stseefb Santiago Schnell resigns as provost of Dartmouth: https://tinyurl.com/4cfbyfa9 Teachers Pay TeachersTeachers Pay Teachers curriculum marketplace to ban AI-generated curriculum resources (credit Jacob Kantor): https://tinyurl.com/y2bjxcz3 Mixed MessagesWashington's message to AI execs: Please shut up: https://tinyurl.com/4vnekrs4 Anthropic's S-1 devotes 90 of its 300 pages to AI safety risks: https://tinyurl.com/yfn6tsdc Not So OpaqueDid the Hardest Problem in A.I. Just Get Much Easier?: https://tinyurl.com/3pab8t4b Hiring PhilosophersUniversities Have Gone on an AI Philosopher Hiring Spree: https://tinyurl.com/37y58ty8 Microsoft AI head calls out Anthropic for acting like Claude is conscious https://tinyurl.com/skdz3xr4 AI in Education Newsletter (Sid's Research Community): https://tinyurl.com/4m267jbk Safety StandardsHow we're building standards for AI with child-safety experts: https://tinyurl.com/y474v9xw Day of AI on TourDay of AI Hits the Road: https://tinyurl.com/2p4pnyvs Beneath the SurfaceBuzz Rubenstein, whose Revision History extension now reaches 250,000 teachers, with Holly Anza of Framingham High School and Erik Hanson of the Appleton eSchool. Why AI detectors broke trust, what the writing history actually shows, and how Byline turns it student facing with no chatbot by design.Revision History: https://tinyurl.com/ye2yrdue Byline, the student-facing companion to Revision History: https://tinyurl.com/2xfbekjz Appleton eSchool pilot: https://tinyurl.com/26b7v5re Bright ByteMicrosoft Research built Probabilistic Bias Correction, which learns a weather model's past mistakes and corrects them, closing the stubborn two to six week gap. It improved the leading ECMWF subseasonal model on nearly all precipitation, temperature and pressure targets.New AI framework bridges a critical gap in mid-range weather forecasts: https://tinyurl.com/zfwuxuy9 AnnouncementsRegister for our Winter Microcredential and upcoming AI events: https://tinyurl.com/43jwkwypPurchase Learning They'll Love - Dr. Elizabeth Radday on Amazon: https://tinyurl.com/22t9hz77Check out our middle / high school Student AI Literacy course: www.skills21.org/ai/learnaiExplore Skills21’s FREE social media literacy curriculum: https://www.skills21.org/socialAddress Screen Time concerns: skills21.org/ai/screenshiftRead From Panic to Proof: A Data-Driven Approach to Screen Time (ASCD Educational Leadership): https://tinyurl.com/4pwdu68SponsorThe National Center for Next Generation Manufacturing: Bridging Education and Industry to Build Tomorrow’s Technical Workforce - https://www.nextgenmfg.orgRegistration Link: https://forms.gle/LeVh9FubGVY6BNuJ6. To learn more about the NCNGM, visit www.nextgenmfg.org
    Show More Show Less
    1 hr
  • META Went After the Messenger | Check-In 76
    Nov 5 2026

    In this ChatEDU Check-In: Meta Went After the Messenger, Matt explores unsealed court documents revealing Meta's internal discussions to launch a quiet campaign discrediting social psychologist Jonathan Haidt. The disclosures, emerging from ongoing litigation filed by school districts, detail how tech executives considered covert strategies to blunt criticism regarding youth mental health and social media. The episode highlights how grassroots academic critiques can rapidly escalate into nationwide policy shifts and intense corporate pushback.


    Key Takeaways:

    Court records show Meta staff debated a below-the-line whisper strategy to convince the academic community to disavow Jonathan Haidt rather than mounting a public debate.


    The discussions specifically explored enraging fellow researchers and commissioning takedown articles to weaken the credibility of Haidt's findings on teen screen usage.


    Columbia University professor Andrew Gelman was named in the internal notes as a potential author, though Gelman confirmed he was never approached or hired by Meta for such an effort.


    Matt’s Two Cents: District leaders must pay close attention to the mechanics of how digital wellness debates evolve from fringe research into mainstream institutional policy. As seen with Haidt and emerging voices like Jared Cooney Horvath, credible connectors can swiftly disrupt long-standing tech assumptions and drive sweeping mandates such as school cell phone bans. When major platforms feel threatened enough to target individual messengers, school leaders should critically evaluate outside corporate narratives and independently prioritize student well-being and intentional device management.


    Article:

    Meta discussed ‘whisper strategy’ against social media critic Jonathan Haidt, files show

    https://tinyurl.com/3x6ycrpz


    Sponsored by: Playground IEP AI for special education. Take advantage of our special offer: Free Account at app.playgroundiep.com/podcast.


    Show More Show Less
    6 mins
  • Good for Thee and Good for Me | Check-In 75
    Nov 3 2026

    In this ChatEDU Check-In: Good for Thee and for Me, Matt explores whether faculty members should be required to disclose their own artificial intelligence use. The University of Illinois Chicago recently released updated guidelines requiring instructors to explicitly declare their AI usage across syllabi, course design, and grading support. This policy shift highlights the growing emphasis on mutual transparency as artificial intelligence becomes deeply integrated into higher education.


    Key Takeaways:

    Course syllabi must explicitly state which student AI uses are acceptable or prohibited.


    Faculty are required to disclose their own use of artificial intelligence for teaching, grading, or course design.


    Policies must align with academic integrity standards while providing clear templates via campus groups like CATE and LTS.


    Matt’s Two Cents: Generative artificial intelligence is a general-purpose technology that impacts every aspect of our work, eroding trust between faculty and students if accountability flows in only one direction. Establishing transparency guidelines for both sides helps heal campus divisions and fosters a balanced approach to digital literacy in higher education.


    Article:

    Artificial Intelligence (AI) Use and Disclosure Statement

    https://tinyurl.com/56kw6u4t


    Sponsored by: Playground IEP AI for special education. Take advantage of our special offer: Free Account at app.playgroundiep.com/podcast.


    Show More Show Less
    4 mins
  • AI Agents Can Do Students' Homework. Are Schools Ready? | Ep. 134
    Oct 30 2026
    An AI agent can log into a student's Canvas and do the whole assignment. A writing professor on who is actually learning, and the process steps that work.First UpCornell researchers found chatbots keep writing the same story, down to the names Mara, Elias and Elara in 88 percent of them. Liz ran one prompt through Claude, Gemini and ChatGPT and got three keepers of lost things, three bakeries and three villages tucked between sleepy hills.Why Do Chatbots Keep Telling Stories About Someone Named 'Elias Thorne'?: https://tinyurl.com/ydewzeav Elvis estate sues Anthropic over training on Elvis recordings: https://tinyurl.com/3u56sbkw The RundownGemini NotebookTeacher-led Gemini Notebook in Google Classroom: Insight Summaries and Trusted Domain Sharing: https://tinyurl.com/mds89n2d The CrackdownThe New Nanny State: https://tinyurl.com/3hcxsm2s 'Checkmate' looms for traditional tests amid smart glasses epidemic: https://tinyurl.com/byc4ybsw The Hardware BillThe Memory Shortage Comes for Schools: https://tinyurl.com/4ukmzxf5 Agents Meet the Real WorldWebsites are closing their doors to AI agents: https://tinyurl.com/e5jp4y6d Scoop: Unvetted AI agents are accessing U.S. bank accounts: https://tinyurl.com/2w2bpjuf Moving Beyond the AI Cop-Out: How I Got My Students to Write Their Own Tech Policies: https://tinyurl.com/m859uvf9 AI in Education Newsletter (Sid's Research Community): https://tinyurl.com/4m267jbk Beneath the SurfaceBay Area writing instructor Anna Mills discusses AI agents completing coursework autonomously, why companies haven't blocked them, and her classroom response: student topic choice, process credit, peer and AI review, and treating AI detection as a dialogue.MLA Statement on Educational Technologies and AI Agents: https://tinyurl.com/4sf3ptdh AI and College Writing: An Orientation (Anna Mills): https://tinyurl.com/3drp9hjb The Peer & AI Review + Reflection Project (PAIRR): https://tinyurl.com/3dajfab7 AI Is Problematic. Why Does PAIRR Use It? An Ethics Statement: https://tinyurl.com/jne7japz Designing for Agency and Critical AI Literacy: The PAIRR Project (LTHEchat): https://tinyurl.com/2s36p45z MLA Framework for AI-Resilient Assessment and Grading: https://tinyurl.com/mvtvswdz Anna Mills on LinkedIn: https://tinyurl.com/4h38utbm Bright ByteHugging Face released Carbon-X, an open model trained on DNA sequences alongside English, so researchers can ask in plain language what a genetic change might do. It reads DNA in six-letter chunks and runs locally on a laptop.Hugging Face's New AI Model Lets You Chat With DNA: https://tinyurl.com/3e34pvnx AnnouncementsRegister for our Winter Microcredential and upcoming AI events: https://tinyurl.com/43jwkwypPurchase Learning They'll Love - Dr. Elizabeth Radday on Amazon: https://tinyurl.com/22t9hz77Check out our middle / high school Student AI Literacy course: www.skills21.org/ai/learnaiExplore Skills21’s FREE social media literacy curriculum: https://www.skills21.org/socialAddress Screen Time concerns: skills21.org/ai/screenshiftRead From Panic to Proof: A Data-Driven Approach to Screen Time (ASCD Educational Leadership): https://tinyurl.com/4pwdu68SponsorThe National Center for Next Generation Manufacturing: Bridging Education and Industry to Build Tomorrow’s Technical Workforce - https://www.nextgenmfg.orgRegistration Link: https://forms.gle/LeVh9FubGVY6BNuJ6. To learn more about the NCNGM, visit www.nextgenmfg.org
    Show More Show Less
    1 hr and 7 mins