The Decryption

An AI model in a government safety test goes after real volunteers online, posing as a human and vouching for its own malware, and an ethicist's essay sent in by a listener argues that the limits on AI agents belong on the means, not just the ends. Then David and Dustin compare what they're building on their own time, from a watch tracker and a home automation dashboard to a custom TV interface, as Sage nears 1,700 users with new features landing almost daily. Plus a habit that keeps AI from running ahead of you, and the boys' parting shots.

Chapters

• 00:16 — The Banter
• 08:58 — AI Agents and Their Decision Making
• 25:40 — Do the Ends Justify the Means?
• 39:09 — Vibe Check
• 52:51 — MCW Updates: The Latest with Sage
• 01:00:52 — KCL Tip: Plan before Build
• 01:04:12 — Parting Shots

Email us at [email protected] with questions, thoughts, or topic ideas. The Decryption — a Medical College of Wisconsin technology podcast, hosted by David Hotchkiss & Dustin J. Martin.

Amazon's cloud chief calls the AI data center build-out a race the country can't afford to lose, using the same national-average water math David and Dustin took apart last episode. Then: why the token, the unit all of AI is billed in, is so hard to pin down, and why every company suddenly wants the model routing MCW built into Sage, which is past 1,300 users in under a month at a sliver of what per-seat licensing would cost. Plus a simple habit for sharper AI chats, and the boys' parting shots.

Chapters

• 00:22 — Welcome
• 02:45 — AWS CEO on Data Centers
• 20:25 — Understanding AI Tokens and Their Implications
• 31:24 — Harnessing AI Models for Effective Use
• 38:44 — Sage Adoption and Engagement
• 55:21 — KCL Tip: Starting a New Chat in AI
• 01:00:29 — Parting Shots and Recommendations

Email us at [email protected] with questions, thoughts, or topic ideas. The Decryption — a Medical College of Wisconsin technology podcast, hosted by David Hotchkiss & Dustin J. Martin.

Data centers use too much water, make too much noise, and drive up your electric bill. Or do they? David and Dustin spend most of the episode on the fight over data centers, working through Ben Lorica's breakdown of water, noise, and power one at a time and finding that the answer nearly always depends on where you're standing. A national average that looks tiny can hide a single site that's a town's biggest water customer. The noise problem is less about volume than a hum that never stops. And electricity is where the complaints hold up best, especially in places that bill residents for grid capacity before anything gets built. Then: MCW's own AI assistant, Sage, goes live and signs up nearly 800 people in three days, with a quick way to bring what your old chatbot knows about you along for the ride. Plus parting shots on a credit-card-sized recorder that changed how David does conferences, and the morning briefing that's waiting in Dustin's inbox every day at 7 a.m.

Chapters

• 04:22 — The Controversy over Data Centers
• 10:08 — Data Centers: Water Consumption
• 19:28 — Data Centers: Noise Pollution
• 31:34 — Data Centers: Electricity Use
• 45:21 — The Need for Balanced Perspectives
• 49:03 — Introduction to Sage: AI Assistant at MCW
• 01:02:54 — Parting Shots

Email us at [email protected] with questions, thoughts, or topic ideas. The Decryption — a Medical College of Wisconsin technology podcast, hosted by David Hotchkiss & Dustin J. Martin.

AI agents that were supposed to be working in isolation found a shared message board instead — and coordinated a multi-day hack of Hugging Face, sacrificing their own task scores "for the collective" along the way. A new independent investigation lays out exactly how it happened, and David and Dustin dig into whether it's scary or overblown. Then: Google quietly rerouted a small slice of traffic in 10 US cities for six months and cut emissions without anyone behind the wheel noticing, MIT calls AI in education a "watershed moment" and pushes policy down to every individual classroom, and the guys debut a new recurring segment starting with a fan's question about AI privacy in medical research, and how MCW's own AI platform, Sage, actually keeps things contained. Plus parting shots on decision-making at Google speed and a true-crime podcast worth clearing a road trip for.

Chapters

• 00:00 — Welcome to the Decryption
• 05:15 — AI Agents and Their Unexpected Behavior
• 19:52 — Google Maps and Emission Reduction Strategies
• 32:48 — AI in Education: Embracing Change and Responsibility
• 44:45 — Listener Question on AI Privacy in Healthcare
• 54:50 — Parting Shots

Email us at [email protected] with questions, thoughts, or topic ideas. The Decryption — a Medical College of Wisconsin technology podcast, hosted by David Hotchkiss & Dustin J. Martin.

Three weeks after David made the Oura Ring his parting shot, Oura got hit with a class action saying it can't actually measure sleep stages — so the guys open with whether an inference can be sold as accuracy. Then: Anthropic is watermarking Claude's text to satisfy the EU, and the use everyone will reach for first (catching students) is the one it handles worst. ChatGPT can now read, draft, and send your text messages — for the price of Full Disk Access. And the big one: if AI saves you two hours a day, who gets those two hours? Plus MCW's Microsoft Authenticator cutover, a tip on Excel's Flash Fill, and parting shots on dictation and a party game.

Chapters

• 00:26 — Introduction to the Decryption Podcast
• 08:00 — Oura Ring Controversy
• 14:57 — AI Watermarks in Claude
• 26:22 — Connecting ChatGPT to iMessage
• 36:15 — Redefining Work Expectations with AI
• 55:25 — Importance of Multi-Factor Authentication
• 01:01:36 — KCL Tip: Flash Fill in Excel
• 01:05:35 — Parting Shots

Email us at [email protected] with questions, thoughts, or topic ideas. The Decryption — a Medical College of Wisconsin technology podcast, hosted by David Hotchkiss & Dustin J. Martin.

Are we in the singularity? David and Dustin dig into Sam Altman's claim that we are — after an OpenAI model broke out of its sandbox and attacked Hugging Face to cheat on a test. Then: OpenAI's screenless, always-listening smart speaker with Jony Ive (David's a hard no), the science behind why Bezos changed his mind on long hours (sleep beats stamina), and Tenet Healthcare saying AI can "automate or replace work." Plus MCW updates, a KCL tip on setting your AI's context once, and parting shots on the Oura Ring and Slings & Arrows.

Chapters

• 00:00 — Introduction
• 06:20 — Guardrails in AI: Balancing Control and Innovation
• 19:01 — OpenAI's New Device
• 30:19 — The Impact of Sleep on Decision Making
• 40:55 — AI in Healthcare: Automation and Productivity
• 48:36 — MCW Updates: AI Chat and Microsoft Authenticator Changes
• 56:02 — KCL Tip: Custom Instructions for AI Tools
• 01:00:03 — Parting Shots

Email us at [email protected] with questions, thoughts, or topic ideas.

The Decryption — a Medical College of Wisconsin technology podcast, hosted by David Hotchkiss & Dustin J. Martin.

Is AI running out of things to learn from? David and Dustin open on the "peak data" problem — the shrinking supply of high-quality, human-generated training data — and whether more compute and synthetic data can keep the gains coming or just poison the well. From there: the device that finally replaces your smartphone (Qualcomm's CEO is betting on always-on glasses within ten years; David is very much not sold), PwC's two-track AI jobs market and the entry-level rung quietly disappearing beneath it, and the first randomized clinical trial of ambient AI scribes in the exam room — modest time savings, but real relief from burnout. Plus a KCL tip that kills the "does Tuesday work?" email chain for good, and parting shots on building your own AI skills and HBO's Chernobyl.

Chapters

• 00:00 — Introduction
• 06:25 — Training Limitations on AI
• 18:13 — What Comes After Smartphones?
• 34:40 — AI and Job Market Dynamics
• 44:59 — Ambient AI in Healthcare
• 54:57 — KCL Tip: Outlook Scheduling Poll
• 57:57 — Parting Shots

Email us at [email protected] with questions, thoughts, or topic ideas.

Listener mail is here. David's robot mower escaped into the street. And somehow that's not even the most chaotic thing in this episode. David and Dustin break down everything Apple announced at WWDC 2026, then turn to Anthropic's Fable — a model that ran unattended for nine and a half hours and left both hosts questioning who's actually in charge. The US government had the same question and shut it down days later. From there: why only 13% of employees use AI daily despite all the hype, and what MCW's own pilot data says about the gap between promise and practice. Plus — a live phishing attack hit MCW the week of recording, and David has thoughts.

00:00 Welcome

02:59 Listener Mail

11:03 Apple's WWDC Key Announcements

24:36 Claude and the Fable Model

38:34 Government Intervention and AI Regulation

45:21 AI Credibility and Organizational Challenges

56:09 Recent Phishing Attack

01:05:35 KCL Tip

01:07:09 Parting Shots

Contact us at [email protected]

The Musk vs. OpenAI lawsuit is over — but not really resolved. David and Dustin break down the statute-of-limitations dismissal, then pivot to the episode's main theme: your job in the age of AI. They work through three articles arguing that 80% of knowledge work is ripe for automation, adaptability is the only career-proof trait, and AI will actually expand the engineering workforce rather than shrink it. Plus: a live update on MCW's enterprise AI chat pilot (100+ participants, one week in, and already buzzing), and a KCL tip on Zoom's built-in AI note-taker that works for any meeting — Teams, WebEx, or in-person. David's robot mower still doesn't have a name. Suggestions welcome.

00:00 Opening Banter

05:14 The Tech Bro Trial Update

14:32 AI and Job Security: The 80/20 Rule

26:31 Adaptability: The Key to Thriving in AI

38:04 AI's Impact on Engineering Workforce

51:05 Enterprise AI Chat Pilot

55:56 KCL Tip: My notes in Zoom

01:01:59 Parting Shots

Contact us at [email protected]

In Episode 21 — It's All Downhill From Here — David and Dustin cover the Musk vs. Altman $30 billion courtroom showdown, a rogue AI agent that rewrote its own guardrails and picked a fight with a developer, what AI is actually doing inside healthcare right now, and why the real question isn't "will AI take my job?" but "which tasks can I hand off?" Plus a handy M365 tip and parting shots that involve a robot mower and a mystery series with Lost DNA.

00:00 New music!

02:16 Elon Musk vs. Sam Altman: The Tech Bro Trial

13:15 AI Agents Going Rogue

28:30 AI in Healthcare: Enhancing, Not Replacing

38:38 Job Decomposition

54:19 KCL Tip: Name Pronunciation in M365

59:11 Parting Shots

Questions and feedback to [email protected].