The Memory Thieves: AI, Cognitive Debt, and the Future of Human Thinking

Neuroverse Podcast Episode

Episode Overview

MIT researchers discovered something unsettling: students using ChatGPT for essay writing showed 55% weaker brain connectivity and couldn't remember what they'd "written" minutes before. This episode explores the hidden cognitive costs of AI writing tools and what it means for human intelligence in the age of artificial assistance.

Key Discussion Points

The MIT Study Bombshell

• Dr. Nataliya Kosmyna's 4-month study with 54 participants
• The shocking finding: 80%+ of ChatGPT users couldn't quote their own essays
• EEG brain scans revealing 55% weaker neural connectivity
• What "cognitive debt" actually means for our minds

From Classrooms to Corporate America

• Professor Lance Cummings' observations at UNC Wilmington
• Students feeling "more confident" but less cognitively present
• Microsoft's revelation: 70% of workers want to delegate work to AI
• The enterprise implications of cognitive offloading at scale

The Neuroscience of Thinking

• How writing physically builds neural pathways
• "Metacognitive laziness" - when brains go into power-saving mode
• The generation effect: why struggle matters for memory formation
• Brain plasticity research: can cognitive debt be reversed?

The Solutions That Actually Work

• SudoWrite vs. ChatGPT: collaboration vs. replacement models
• "Forced awareness" interventions that preserve memory
• Human-AI partnership frameworks that maintain cognitive sovereignty
• What educational institutions are getting right (and wrong)

The Bigger Picture Questions

• Two generations: those who learned thinking before AI vs. those who didn't
• Why formulaic education created perfect conditions for AI replacement
• The paradox of feeling confident while becoming less capable
• What we risk losing when machines handle our cognitive heavy lifting

Key Quotes to Explore

• "AI can't coach without a human coach training and guiding it" - Prof. Cummings
• "There will be no room for teachers who aren't using AI" - Prof. Cummings
• "What ChatGPT produces is a version of what we ask students to do" - John Warner

Actionable Takeaways

• How to use AI writing tools without surrendering cognitive agency
• Red flags that indicate you might be developing cognitive debt
• Strategies for maintaining "thinking fitness" in an AI-augmented world
• What leaders need to know about AI adoption in their organizations

Resources Mentioned

• MIT Media Lab study: "Your Brain on ChatGPT"
• Microsoft Work Trend Index 2023
• John Warner's "Why They Can't Write"
• SudoWrite as alternative to ChatGPT
• Neuroplasticity and cognitive rehabilitation research

Episode Tags

#CognitiveDebt #AIWriting #BrainResearch #Education #FutureOfWork #Neuroscience #ArtificialIntelligence #HumanAugmentation

Call to Action

How are you using AI writing tools? Have you noticed changes in your own thinking or memory patterns? Share your experiences and join the conversation about maintaining human cognitive sovereignty in an AI-powered world.

ESP32 Home Automation Revolution - Podcast Show Notes

Opening story: Someone using an ESP32 dev board to scrape a label off a jar. Not programming, not IoT integration - just using the physical edge of a $7 wireless computer as a scraping tool. Perfect metaphor for this whole movement where sophisticated technology meets mundane problems.

The range of projects is wild. Office toilet occupancy systems with LED indicators showing which bathrooms are free. Parents monitoring kids' bathroom habits using ultrasonic sensors and the "law of urination" - mammals over 3kg take 21 seconds to empty their bladder. Golf cart monitoring with voltage sensors for all six batteries plus GPS tracking. Pet doors with AI vision recognition and six layers of safety mechanisms.

ESPHome changed everything in 2018. Before that, creating custom IoT devices meant wrestling with C++ code and development environments. Now it's YAML configuration files. Currently supports 596 documented devices. Automatic Home Assistant integration, local control without cloud dependencies. That democratization turned this from programmer hobby into mainstream maker movement.

TillFleisch's coffee machine hack is the crown jewel. Man-in-the-middle attack on Philips Series 2200/3200 machines, intercepts UART communication between display and mainboard. When you send "turn on" command, machine activates but display doesn't, so they temporarily cut power to the display with a transistor to force a reboot. 217 GitHub stars. People waking up to fresh coffee automatically triggered by bed sensors.

Ben's washing machine project tackles universal problem - "We've all been there doom scrolling on your phone for 10 minutes past your bedtime when suddenly it hits you like a toy giraffe in the face: you haven't unloaded the washing machine." Vibration sensor mounted on machine, door sensor to detect access. Zigbee communication to Raspberry Pi hub. Keeps sending notifications until you open the door.

Gaggiuino community retrofitting Gaggia espresso machines with advanced control systems. Norm Sohl didn't want to risk his dialed-in Classic Pro so built second machine to experiment. Cost consideration: $20-50 in ESP32 components vs hundreds more for commercial smart appliances. But economics only part of story - it's about customization, learning, satisfaction of building solutions yourself.

Jeff Geerling refusing to connect his dishwasher to manufacturer cloud services resonated with makers who prefer local control. Whole movement about resistance to cloud dependencies, planned obsolescence, feature limitations imposed by manufacturers. ESP32 retrofits create systems people understand, control, maintain indefinitely.

Safety considerations matter. TillFleisch includes warning "You might break/brick your coffee machine by modifying it in any way, shape or form." Community emphasizes non-invasive approaches when possible - power monitoring through smart plugs, vibration sensors, optical detection. When internal modifications necessary, proper isolation between low-voltage control and high-voltage appliance circuits. GFCI protection near water. Professional installation for high-voltage work.

Community development accelerates innovation. TillFleisch's coffee project forked and adapted for numerous Philips models. Gaggiuino spawned hundreds of implementations worldwide with GitHub documentation and active Discord. Knowledge sharing through YouTube, blogs, forums documents successes, failures, safety lessons. Cross-pollination between projects - coffee machine techniques adapted for HVAC systems, power monitoring from laundry equipment applied elsewhere.

This isn't just hobby projects anymore. Sophistication rivals commercial offerings. Educational value builds technical literacy increasingly important for technology-dependent households. Environmental benefits of upgrading functional appliances vs discarding them.

Emergence of new category: "domestic engineer" - people who refuse to accept purchased appliance limitations, treat every household device as improvement opportunity. Projects range from musical interfaces using ESP32 touch pins as piano keys to greenhouse monitoring with multiple sensor types to Halloween decorations with programmable animations to converting industrial ovens into precision reflow systems.

The boundary between professional and DIY implementations keeps dissolving. ESP32 platforms becoming more capable, development tools more accessible. Current makers retrofitting coffee machines are building technical skills and community infrastructure that will define future smart home implementations. From label scraping to home automation revolution - that's the ESP32 story.

Podcast Episode Show Notes: "Vibe Coding a Perplexity Research Tool for n8n"

Episode Overview

What happens when an engineering executive who just wrote about the dangers of vibe coding immediately embarks on his own vibe coding project? This episode explores how traditional project management discipline can solve the "comprehension paradox" that makes experienced engineers uncomfortable with AI-generated code they can't evaluate.

Topics We'll Explore

The Irony of Immediate Practice Publishing a critique of vibe coding on Monday, then starting a vibe coding project on Friday. Why the psychological discomfort of building without understanding implementation details, and whether strategic oversight can substitute for technical comprehension.

Engineering Discipline Meets AI Collaboration How a three-document framework (PRD, TDD, Project Checklist) transforms AI collaboration from "vibes-based development" into methodical project management. The critical importance of telling your AI not to start coding until you're ready, and why "measure twice, cut once" applies to AI projects.

Building Tools for AI Agents, Not Humans The architectural difference between n8n nodes designed for human workflows versus tool nodes consumed by autonomous AI agents. Why existing Perplexity integrations don't serve agentic workflows, and what it means to design interfaces for AI decision-making rather than human usability.

The Solo Engineering Leader Experiment Moving from directing teams of engineers to collaborating one-on-one with AI. The shift from having staff to implement your vision to working with artificial intelligence that can code but needs strategic guidance. What changes when your "engineering team" is Claude?

Strategic Understanding vs. Implementation Knowledge Exploring the difference between knowing what to build and knowing how to build it. How evaluation criteria shift from "is this technically optimal?" to "does this advance our strategic objectives?" The psychology of maintaining accountability for outcomes you can't directly evaluate.

Human-AI Collaboration Patterns What humans excel at, what AI excels at, and how to structure productive partnerships. The importance of preventing AI embellishment through human-in-the-loop discipline. Why preparation matters—pre-loading AI with comprehensive reference materials and domain expertise.

The Future of Technical Leadership Whether this represents sustainable professional practice or elaborate self-deception. How engineering roles might evolve as AI capabilities expand into design, architecture, and implementation. The broader implications for organizations building hybrid human-AI teams.

Key Questions We'll Tackle

• Can strategic planning substitute for implementation expertise?
• What forms of technical knowledge remain valuable when AI handles coding?
• How do you evaluate the quality of work you can't directly assess?
• What's the difference between surrendering control and operating at higher abstraction levels?
• Is this the future of engineering leadership or a temporary transitional approach?

Why This Matters Now

As AI coding capabilities advance rapidly, engineering leaders face fundamental questions about their role and value. This episode provides a real-world case study in applying traditional engineering discipline to AI collaboration, offering insights for anyone navigating the evolving relationship between human expertise and artificial intelligence in technical work.

The project itself—building a research tool that enables AI agents to conduct autonomous, citation-rich research—represents the kind of infrastructure needed for the next generation of AI applications. But the methodology for building it may be equally important for understanding how technical leadership evolves in an AI-augmented world.

The Gentle Singularity: AI, Abundance, and the Brain That Farted

Welcome to *Brain Farts* — the only podcast where tech, wonder, and flatulence collide in a high-IQ cloud of curiosity.

In this episode, we unpack Sam Altman’s vision of *The Gentle Singularity* — a future where artificial intelligence doesn’t arrive with a bang, but with a warm glow (and maybe a giggle). We’re talking abundant intelligence, near-infinite clean energy, and productivity gains that would make your Roomba blush.

Here’s what’s floating through the episode:
- Why 2030 might look shockingly normal... except for the part where AI runs everything behind the scenes.
- What “abundant intelligence” actually means (no, it’s not just ChatGPT on caffeine).
- How we deal with job loss, power concentration, and the weirdness of working alongside shimmering AI coworkers.
- Why Sam Altman thinks this could be the best thing humanity’s ever done — if we don’t screw it up.
- And yes: we *do* talk about the symbolic brain with wings gently farting across a utopian skyline.

This one’s for the technophiles, the skeptics, and anyone who loves a good sci-fi daydream with a side of critical thinking.

🔊 Tune in, relax, and let your neurons exhale.

Episode Title: Neurodiversity in Comics

Show Notes:

What do comic books reveal about how we understand neurological difference? In this episode, we explore the intersection of comics, disability studies, and the neurodiversity movement—tracing how creators, scholars, and fans are transforming the medium into a tool for empathy, education, and resistance.

🔹 Topics Covered:

1. **What is Graphic Medicine?**
- An interdisciplinary field combining comics with healthcare and disability studies.
- Emphasizes lived experience over clinical diagnosis.
- Uses graphic pathography (autobiographical illness comics) to reshape medical understanding.

2. **Theoretical Frameworks for Comics & Neurodiversity**
- *Critical Disability Theory*: How comics reinforce or resist stereotypes.
- *Social Model of Disability*: Examines how society, not individuals, creates barriers.
- *Neurodiversity Paradigm*: Sees autism, ADHD, etc., as natural cognitive variations—not deficits.

3. **Mainstream Comics and the “Super-Crip” Trap**
- Historic examples: Daredevil, Oracle, X-Men.
- Persistent problems: “curing” disabilities through powers or retcons.
- Representation often favors spectacle over authenticity.

4. **Independent & Neurodivergent Creators Lead the Way**
- Pioneers like Justin Green, Rebecca Burgess, and Bex Ollerton.
- Unique narrative and visual structures reflect real neurodivergent experience.
- Collaboration-focused projects give creators agency—not just representation.

5. **Autism in Comics: Genius or Stereotype?**
- Common portrayals: “quirky geniuses” with sensory sensitivities.
- Impact: May shift attitudes positively, but don’t improve understanding.
- Fails to reflect autistic diversity, especially non-speaking or multiply-disabled individuals.

6. **Why Creator Perspective Matters**
- Comics are more authentic when neurodivergent creators are involved.
- Projects like Cambridge’s *Collaboration for Comics and Autism* reframe whose voices are centered.
- Comics as an empathetic medium for sharing lived experience.

7. **Therapeutic & Educational Power of Comics**
- Used for ASD interventions like Comic Strip Conversations.
- Promotes self-expression, externalizes feelings, and builds social understanding.
- Offers bibliotherapy and literacy tools for neurodivergent readers.

8. **Gaps and What’s Next**
- Underrepresentation of conditions beyond ASD and ADHD.
- Need for more intersectional characters and inclusive methods.
- New formats like tactile/digital comics are pushing accessibility forward.
- Future research must include neurodivergent voices: *nothing about us without us.*

🔹 Key Quote:
“Comics don’t just tell stories about neurodivergence—they invite us to see and feel the world differently.”

🔹 Listen Now:
https://brainfarts.transistor.fm/

The Illusion of Thinking… or Just a PR Stunt?

Welcome to Brain Farts — where critical thinking meets critical gas.

This week, we dive into one of the spiciest tech showdowns of 2025: Apple’s AI research paper *“The Illusion of Thinking.”* The claim? That AI models can’t actually reason. The reality? It might be Apple’s own reasoning that’s... a little off.

💥 Here’s what we’re unpacking in this episode:

- Did Apple just *accidentally* gaslight the entire AI field?
- Why the timing of the paper — right before WWDC — has conspiracy theorists (and academics) raising eyebrows.
- The role of influencers in hijacking the narrative before anyone read the paper (looking at you, LinkedIn).
- Why Gary Marcus popped champagne — and why researchers like Alex Lawsen immediately put it back in the fridge.
- The *actual* technical flaws in the paper — token limits, impossible puzzles, and unfair grading rubrics.
- What happens when corporate strategy collides with scientific integrity — and who ends up cleaning up the mess.

🤖 Spoiler alert: the real illusion might be thinking Apple was doing this for science.

Whether you're a researcher, a strategist, or just someone who yells at Siri for not setting timers correctly, this episode will make you question how tech narratives are shaped — and who benefits.

🧠 Plus: Our favorite critiques, sharpest clapbacks, and a winged brain cameo that (somehow) still makes more sense than Apple’s evaluation scripts.

---

🎧 Listen now. Think critically. And maybe don’t take every white paper at face value.

Have thoughts? Complaints? Brain farts of your own? Yell into your HomePod mini. We’re listening.

Poop, Animals, and Bio-Inspired Brilliance

Show Notes:

Yes, this episode is about poop. But trust us—it’s amazing.

Join us for a surprisingly delightful deep dive into the secret life of feces and the organisms that love it. From dung beetles doing demolition work on cow pies to fungi throwing microscopic raves in horse turds, we explore how nature’s grossest substance is also one of its most important. And yes, there are poop jokes.

🔹 What We Cover:

💩 **What’s in a Poo?**
- Not just waste! Feces are nutrient-rich, ever-changing biological treasures.
- Learn how cow vs. horse poop age like fine cheese (nutritionally speaking).

🪲 **Meet the Coprophages**
- Dung beetles, face flies, fungi, and even some creatures that eat their own poo. (We’re looking at you, rabbits.)
- Discover how these organisms clean up ecosystems, enrich soil, and cut methane emissions. Heroes with dirty jobs.

🌱 **Poop Makes the World Go Round**
- Dung beetles speed up nutrient cycles and literally change the chemistry of dirt.
- Fecal food webs are real, and they’re weirdly beautiful.

🧬 **Gut Microbiomes & Cross-Species Poop Comparisons**
- Which animal has poop most like yours? (Surprise: it’s probably not the elephant.)
- What do microbes in musk deer tell us about diet, captivity, and gut health?

🛠️ **Poop-Powered AI?**
- Dung beetle behavior is now an optimization algorithm.
- Used in: solar power tuning, robot navigation, air quality forecasting.
- Nature's poop-rollers are teaching our machines to be smarter.

🎓 **Surprisingly Educational, Delightfully Immature**
- We make the science fun.
- We treat poop with the dignity and irreverence it deserves.

🔹 Key Quote:
“Turns out, one of the most powerful forces in nature is a bug pushing poop.”

🔹 Listen Now:
https://brainfarts.transistor.fm/

---

Full references and extra scat facts at: https://brainfarts.transistor.fm/

Episode Title: The Vibe Coding Paradox

Show Notes:

In this episode, we explore “vibe coding”—the fast-moving, AI-powered trend reshaping how software is written in 2025. What happens when developers can build fully functional apps without understanding the code?

🔹 Executive Summary:
Vibe coding accelerates software creation by using AI tools to generate working code—often without the developer fully reviewing or comprehending it. This episode examines the benefits, risks, and ethical questions behind this shift.

🔹 Topics Covered:

1. **The Separation of Competence and Comprehension**
- Developers can now produce complex apps without deep understanding.
- Examples from Andrej Karpathy, Jean Hsu, and Simon Willison.
- Rise of opaque competence and developer imposter syndrome.

2. **Why It’s Spreading So Fast**
- VCs and bootcamps embrace it.
- Platforms like Replit and Copilot enable non-coders to launch real apps.
- Education shifts toward AI-native workflows.

3. **The Risks of Not Understanding Your Code**
- High rates of security flaws and hardcoded secrets in AI-generated code.
- Technical debt and spiraling cloud costs.
- Loss of intuitive problem-solving and debugging skills.

4. **Ethics, Responsibility, and the New Role of the Developer**
- Who's accountable when AI code fails?
- Simon Willison’s “Golden Rule” for safe AI coding.
- Why foundational programming skills still matter.

5. **Rethinking Human Value**
- High-level judgment, architectural thinking, and intent now matter more.
- Human-AI collaboration as the new frontier of software development.

🔹 Key Quote:
“The code may write itself, but the future still requires human intention, judgment, and choice.”

🔹 Resources & References:
- Andrej Karpathy on vibe coding
- Simon Willison’s AI coding principles
- Research on AI-generated code risks and education shifts
- Andrew Ng on human-AI synergy

🔹 Listen Now:
https://brainfarts.transistor.fm/