Tech Unfiltered with Dr. Mike

Tech Unfiltered with Dr. Mike

By Dr. Michael LitmanNewsTechnologyTech News
Download on the App Store

All Tech Unfiltered with Dr. Mike episodes

  • Claude-Assisted Riemann Proof: What Independent Checks Establish

    The Lean and Mathlib communities say they have independently verified both the statement and the formal proof of the Riemann Hypothesis developed by mathematician Levent Alpöge with Claude. Dr. Mike explains the November 21 verification statement, why this question matters for prime numbers, and what still separates a checked proof from a clear human explanation and formal prize recognition.

    November 21, 2026 edition of Tech Unfiltered with Dr. Mike.

    Takeaways:
    • The joint statement reports four checks covering the mathematical statement, reproducible compilation and export, statement matching, and independent proof checking. This episode reports their verification; it does not claim to have independently verified the proof itself.
    • Alpöge describes human mathematical guidance and Claude's contributions, and plans to simplify and explain the argument.
    • The Riemann Hypothesis concerns how prime numbers deviate from their average distribution. Understanding the proof's ideas remains an important next step.
    • Clay's prize process is separate: qualifying publication, at least two years after publication, and general acceptance in the mathematics community. This episode does not claim a prize has been awarded.
    • Mike's interpretation: confidence should stay attached to the particular claim that was checked. A clearly labeled hypothetical example explains why statement matching matters; the joint statement says this proof passed that check.
    • No immediate consumer benefit, broken encryption or generally reliable AI system follows from this result alone.

    Chapters:
    00:00 Show intro
    00:10 The Riemann proof
    00:51 Four independent checks
    01:32 Human and AI contributions
    01:58 Why prime numbers matter
    02:31 Understanding comes next
    02:51 Prize recognition
    03:19 What verification establishes
    03:44 Checking the right question
    04:04 What to watch next

    Sources:
    Lean and Mathlib joint verification statement (November 21): https://leanprover-community.github.io/blog/posts/riemann-hypothesis/
    Levent Alpöge's author note (November 20; background): https://alpo.ge/remark.pdf
    MIT, Andrew Sutherland verification status (November 21 update): https://math.mit.edu/~drew/grh/
    Clay Mathematics Institute, Riemann Hypothesis background: https://www.claymath.org/millennium/riemann-hypothesis/
    Clay Mathematics Institute, Millennium Prize rules: https://www.claymath.org/millennium-problems/rules/
    Le Monde, independent reporting (November 21): https://www.lemonde.fr/en/science/article/2026/11/21/the-likely-solution-to-the-riemann-hypothesis-by-ai-marks-a-turning-point-for-mathematics_6786232_10.html

    This episode uses AI-generated narration and editorial illustrations. The visuals are explanatory scenes, not photographs of the researchers or reproductions of the actual proof. The standard show intro remains at normal speed; the episode video after the intro is presented at 1.10×. The RSS audio edition retains normal-speed narration.

    This audio edition uses AI-generated narration.

    Watch the illustrated YouTube edition

    6 min
  • Trump Proposes Tax on AI Data Centers Abroad: What Is Still Unknown

    President Trump has threatened a tax on American companies that build AI data centers abroad. Dr. Mike examines what was announced, what remains unspecified, and why communities should look closely at the written details of any proposed facility.

    November 20, 2026 edition of Tech Unfiltered with Dr. Mike.

    Takeaways:
    • The announcement is a stated intention. The reporting reviewed here does not establish an operating ban or specify a tax rate, start date, or implementation mechanism.
    • Separate CNBC remarks by Treasury Secretary Scott Bessent concern access to computing. Location and access are related questions, but the remarks do not fill in the proposed tax's missing details.
    • Mike's interpretation: announcements, formal rules, and individual projects require different evidence.
    • A hypothetical town illustrates questions about electricity, water, noise, construction work, continuing jobs, and enforceable commitments. These examples are not findings about a particular facility.
    • Household benefits and burdens remain questions to investigate. The episode does not predict bills, job totals, or tax revenue.

    Chapters:
    00:00 Show intro
    00:10 AI data centers abroad
    00:32 The announced tax threat
    01:06 Rate, timing and implementation
    01:34 Location and remote access
    01:58 Announcement, rule and project
    02:29 Questions for a town
    03:01 Written commitments and jobs
    03:30 Benefits and burdens
    04:00 What to watch next

    Sources (published November 20, 2026):
    Associated Press: https://apnews.com/article/f1ac03b73d13445400eeed8631d9d879
    Just the News, Ben Whedon: https://justthenews.com/nation/technology/trump-announces-ban-companies-building-data-centers-outside-us
    CNBC official interview transcript, Treasury Secretary Scott Bessent (separate context): https://pressroom.versantmedia.com/cnbc/press-releases/cnbc-exclusive-transcript-us-treasury-secretary-scott-bessent-speaks-cnbcs-1

    This episode uses AI-generated narration and editorial illustrations. The visuals are explanatory illustrations, not photographs of an actual policy announcement or specific facility. The standard show intro remains at normal speed; the episode video after the intro is presented at 1.10×. The RSS audio edition retains normal-speed narration.

    This audio edition uses AI-generated narration.

    Watch the illustrated YouTube edition

    6 min
  • AI Used in European Commission Cyberattack: What Officials Confirmed

    The European Commission has confirmed that AI tools were used in a cyber campaign against it. Dr. Mike separates the official account, independent reporting, and unanswered questions about disruption, personal information, and the role of AI.

    November 19, 2026 edition of Tech Unfiltered with Dr. Mike. The incident was detected on October 22; November 19 is the date of public confirmation, not the start of the attack.

    Takeaways:
    • Several applications were deliberately taken offline as a protective measure. Most have been restored, while the investigation continues.
    • Officials did not identify the attacker, name the AI tool, characterize the scale, or conclusively declare the campaign over.
    • The statement that investigators had not detected a breach affecting staff as a whole is not an assurance that every individual is unaffected. Uncertainty also does not prove that information was stolen.
    • The Record reports personnel-service disruption. Availability and confidentiality are separate questions; restoration does not settle every question about exposure.
    • Updating software, using unique passwords and multifactor authentication, and checking official support channels are general precautions, not findings about the cause of this attack.

    Chapters:
    00:00 Show intro
    00:10 AI in a cyber campaign
    00:57 Applications taken offline
    01:28 What is known about personal information
    01:55 What independent reporting adds
    02:20 What AI use does not establish
    02:50 When everyday systems stop
    03:15 Availability and confidentiality
    03:43 Useful account precautions
    04:08 Facts and unanswered questions

    Sources (November 19, 2026):
    European Commission midday briefing (cyberattack discussion begins at 08:46): https://audiovisual.ec.europa.eu/en/media/video/I-298222
    The Record, Alexander Martin: https://therecord.media/european-commission-confirms-cyberattack-using-ai
    Reuters, republished by Euronext: https://live.euronext.com/en/financial-news/eu-commission-confirms-it-was-targeted-cyberattack-through-use-ai-tools

    This episode uses AI-generated narration and editorial illustrations. The visuals are explanatory illustrations, not images of the actual cyber campaign. The standard show intro remains at normal speed; the episode video after the intro is presented at 1.10×. The RSS audio edition retains normal-speed narration.

    This audio edition uses AI-generated narration.

    Watch the illustrated YouTube edition

    6 min
  • Gemini Tasks Goes Free: What You're Giving AI Permission to Do

    Google is expanding free access to Tasks in Gemini, an assistant that can work on multi-step projects and recurring routines. Dr. Mike explains the rollout, what connected features can do, and why purchase permissions deserve a careful look.

    November 18, 2026 edition of Tech Unfiltered with Dr. Mike. Google announced the expansion on November 17; Android Central reported on it November 18.

    Takeaways:
    • Free access is rolling out to adults in the United States, Japan, South Korea, Canada, Australia and India. Availability may vary, and some advanced features begin in the U.S.
    • Google describes purchase confirmation or permission granted in advance, including an always-allow option for routine actions. A fresh approval is not promised for every purchase.
    • Google’s safeguards are company claims, not an independent security audit. Errors, unfinished tasks and website handoffs remain possible.
    • Start with a clearly limited request, review connected accounts and permissions, and check the actual confirmation or receipt. The oil-change example is hypothetical; this episode is not a report of personal product testing.

    Chapters:
    00:00 Show intro
    00:10 Gemini Tasks goes free
    00:55 Free access in six countries
    01:21 Tasks that can take action
    01:46 Purchase permissions matter
    02:20 Mistakes and unfinished tasks
    02:48 A booking has consequences
    03:40 Give a clear, limited request
    04:09 Free access still needs judgment

    Sources:
    Android Central (November 18, 2026): https://www.androidcentral.com/apps-software/ai/google-just-made-its-best-gemini-agent-free-heres-where-you-can-get-it
    Google announcement (November 17, 2026): https://blog.google/innovation-and-ai/products/gemini-app/tasks-in-gemini-free-new-features/
    Google product and eligibility information: https://gemini.google/overview/productivity/

    This episode uses AI-generated narration and editorial illustrations. Visuals illustrate the discussion and are not screenshots of the product. The standard show intro remains at normal speed; the episode video after the intro is presented at 1.10×. The RSS audio edition retains normal-speed narration.

    This audio edition uses AI-generated narration.

    Watch the illustrated YouTube edition

    6 min
  • Nashville's AI Traffic Signals: What the Expansion Really Means

    What can lidar and artificial intelligence tell us about a busy intersection? In this November 17, 2026 edition of Tech Unfiltered with Dr. Mike, we examine Nashville’s announced traffic-sensing expansion and the questions that matter for people walking, cycling, and driving.

    Ouster says its BlueCity system was selected for 100 additional intersections, bringing its total contracted Nashville deployments to more than 200 intersections and roadway locations. Contracted does not mean all locations are installed or operating today. Smoother traffic and improved safety are stated aims, not demonstrated results from this new expansion.

    Takeaways:
    • Lidar uses laser ranging to build a spatial model; measurements are not human understanding.
    • City documents describe counts, routes, and near-miss analysis. Those background documents do not establish results for today’s contract.
    • The crossing example is hypothetical, not a verified citywide feature.
    • Mike’s questions concern testing, fallback procedures, public outcomes, privacy, and accountability; they are not allegations of known defects or abuse.
    • Keep following posted signals and normal safe crossing practices. Never assume a sensor has detected you.

    SOURCES
    Ouster announcement (November 17, 2026; company claims)
    https://investors.ouster.com/news-releases/news-release-details/ouster-bluecity-rev8-selected-second-multimillion-dollar-traffic
    Axios reporting (November 17, 2026)
    https://www.axios.com/2026/11/17/traffic-lights-smart-lidar-ai-nashville-ouster
    Nashville LADDMS program (city background)
    https://www.nashville.gov/departments/transportation/traffic-management-center/laddms-program
    Nashville lidar trailer information (September 2025 background)
    https://www.nashville.gov/sites/default/files/2025-10/NDOT_LiDAR_Trailer_Informational.pdf?ct=1761693111

    This episode uses AI-generated narration and editorial illustrations. The illustrations are explanatory scenes, not documentary images, real system interfaces, or evidence of deployed features. Reported facts, company claims, city background, and Mike’s interpretations are distinguished throughout.

    This audio edition uses AI-generated narration.

    Watch the illustrated YouTube edition

    6 min
  • FAA's SMART AI and the Ground Stop: What the Evidence Says

    What does the FAA’s SMART system actually do, and what does the evidence say about the November 13 ground stop? In this November 16, 2026 edition of Tech Unfiltered with Dr. Mike, we separate the FAA’s official account, new Associated Press reporting, early agency results, and unresolved oversight questions.

    The FAA says SMART did not cause the ground stop and was unrelated to the affected network. Its November 15 account describes excessive connection attempts affecting legacy telecommunications, with root-cause investigation continuing. That is the agency’s account, not a completed independent forensic finding.

    Takeaways:
    • SMART supports air-traffic planning; people retain operational decisions and aircraft-separation responsibilities.
    • The FAA’s November 10 expansion announcement covered all 22 air-route traffic control centers.
    • Reported time and fuel savings are early agency results, not independent safety proof or a guarantee for your flight.
    • An October 13 congressional audit request raises questions; it is not an audit finding or proof of an accident.
    • Training, time to evaluate recommendations, authority to disagree, and fallback procedures are Mike’s recommendations, not claims about undocumented FAA procedures.
    • Check your airline and official FAA information for current flight status.

    SOURCES
    Associated Press via ABC News (new reporting, November 16)
    https://abcnews.com/US/wireStory/questions-swirl-faas-new-smart-ai-system-officials-138170875
    FAA statements (November 10 expansion announcement; background)
    https://www.faa.gov/newsroom/statements/general-statements
    FAA ground-stop questions and answers (November 15; official account)
    https://www.faa.gov/newsroom/nationwide-ground-stop-questions-and-answers
    FAA SMART fact sheet (background)
    https://www.faa.gov/newsroom/SMART_One-Pager.pdf
    House Transportation Democrats (October 13 audit request; background)
    https://democrats-transportation.house.gov/news/press-releases/ranking-members-larsen-carson-tandi-democrats-request-audit-of-faas-ai-based-air-traffic-software

    This episode uses AI-generated narration and editorial illustrations. Illustrations explain the topic; they are not documentary images of the ground stop, real operational screens, or evidence of autonomous aircraft control. Editorial facts and sources were reviewed separately from interpretation and recommendations.

    This audio edition uses AI-generated narration.

    Watch the illustrated YouTube edition

    6 min
  • AI Exam Cheating in South Korea: What Does a Score Really Prove?

    South Korea's top tax-accountant exam scorer has admitted using a miniature camera, a phone and generative AI to cheat. In this November 15, 2026 edition of Tech Unfiltered with Dr. Mike, we examine the reported admission, the government's response, and what a credential should tell students, employers and clients.

    The reported 159-of-160 first-round score is an exam result, not an AI benchmark. Reporting does not establish the model used or prove independent professional competence. Police are investigating; an admission is not a completed court judgment.

    Takeaways:
    • Separate reported facts, announced government plans and Dr. Mike's interpretation.
    • Protect honest candidates, legitimate accommodations and fair process.
    • Distinguish permitted AI learning tools from unauthorized exam assistance.
    • Make tool rules clear and ask people to explain their reasoning. These are Mike's recommendations, not adopted South Korean policy.

    SOURCES
    South Korean Ministry of Employment and Labor (Nov. 15 statement)
    https://moel.go.kr/news/enews/report/enewsView.do?news_seq=20256
    Yonhap (Nov. 15)
    https://en.yna.co.kr/view/AEN20261115000753315
    The Korea Herald (Nov. 15)
    https://m.koreaherald.com/article/10937538
    The Korea Times (updated Nov. 15 KST)
    https://www.koreatimes.co.kr/southkorea/law-crime/20261115/top-tax-accountant-exam-scorer-admits-to-cheating-using-ai

    This episode uses AI-generated narration and editorial illustrations. Illustrations are explanatory reconstructions, not documentary images of the candidate or events. Editorial facts and sources were reviewed separately from opinion and recommendations.

    This audio edition uses AI-generated narration.

    Watch the illustrated YouTube edition

    6 min
  • ChatGPT Wallet: Who Approves the Purchase?

    November 14, 2026 edition. ChatGPT Wallet lets eligible U.S. Pro users have an assistant complete checkout after they review and approve each purchase. Dr. Mike follows a hypothetical refrigerator-filter order to explain card privacy, order accuracy, and who handles a return or billing problem.

    TAKEAWAYS
    • The initial rollout is limited, and ordinary ChatGPT chats do not provide wallet access.
    • Substitute card details protect the actual card number; users still need to check the item, seller, total and recurring charges.
    • Merchants handle fulfillment and returns. Card issuers handle billing disputes under existing card terms.
    • Advance purchase permissions are a future proposal, not the current approval process.

    SOURCES
    https://openai.com/index/introducing-chatgpt-wallet/
    https://decrypt.co/383278/openai-chatgpt-wallet-ai-agents-shop-without-crypto
    https://www.ndtv.com/feature/chatgpt-wallet-can-now-shop-and-pay-for-you-how-openais-new-wallet-works-12325673
    https://openai.com/policies/chatgpt-wallet-terms/

    This episode uses AI-generated narration and editorial illustrations. Shopping scenes are hypothetical examples, not screenshots of the product.

    This audio edition uses AI-generated narration.

    Watch the illustrated YouTube edition

    6 min
  • When Copilot Loses the Inbox: What the Microsoft 365 Outage Changed

    2026-11-13 edition. Microsoft’s email outage left some people waiting for messages and AI assistants unable to reach workplace data. Dr. Mike follows a hypothetical rescheduled appointment to explain how an accurate summary of an old invitation can still send you to a meeting at the wrong time.

    TAKEAWAYS
    • Which Copilot functions depend on access to mail, calendars and files.
    • Why switching assistants can still leave you relying on the same mailbox.
    • How to check the time on a source message and confirm an urgent change by phone.

    SOURCES
    https://mailservices.isc.upenn.edu/computing/email/penno365/alerts/ms-incidents.html
    https://apnews.com/article/microsoft-outage-outlook-email-365-status-down-d92547a821930290db2964388016979a
    https://english.elpais.com/technology/2026-11-13/microsoft-fails-to-restore-services-after-more-than-12-hours-of-global-disruption.html

    This episode uses AI-generated narration and editorial illustrations. The artwork is conceptual, not a record of actual events.

    This audio edition uses AI-generated narration.

    Watch the illustrated YouTube edition

    6 min
  • Can a Court Pause the Next AI Model? Four Groups Sue Five Developers

    2026-11-12 edition. Four nonprofits asked a California judge to freeze development and release of more capable AI models. Dr. Mike walks through what the groups requested, why the case could matter to people who never use a chatbot, and what filing a lawsuit changes today.

    TAKEAWAYS
    • The request covers both building and releasing more capable systems.
    • The plaintiffs also challenge letting AI direct research into its successors.
    • A request for an order leaves the court’s decision open.

    SOURCES
    https://cdn01.dailycaller.com/wp-content/uploads/2026/11/EMBARGOED_%20Alliance%20Lawsuit%20Press%20Release%20Draft.pdf
    https://apnews.com/article/artificial-intelligence-ai-lawsuit-cyberattacks-anthropic-openai-03ece37b96979fe834f60976648504ba
    https://kwsn.com/2026/11/12/openai-google-anthropic-hit-with-lawsuit-seeking-halt-to-frontier-ai-development/

    This episode uses AI-generated narration and editorial illustrations. The artwork is conceptual, not a record of actual events.

    This audio edition uses AI-generated narration.

    Watch the illustrated YouTube edition

    6 min