The AI Fix

In episode 98 of The AI Fix, guest host Nik Roberts explains the cognitive debt you're racking up every time you let ChatGPT do your thinking, and Mark explains the new eval that's got even the very best AI models completely stumped.
Also in this episode, Nik learns how to pronounce "Mythos" and Mark learns how to pronounce "Kasparov"; Gen Z fights back against AI; Britain's AI Security Institute tests Mythos; DeepSeek casually drops a 1.6 trillion parameter model; and a Sony robot beats the elite at the world's most boring pastime.
Episode links:
Gen Z Sabotaging AI at Work So It Won't Take Their Job.
Our evaluation of Claude Mythos Preview's cyber capabilities.
DeepSeek previews new AI model that 'closes the gap' with frontier models.
Inside Project Ace: Discover the Robot Athlete That Competes With Professional Table Tennis Players.
Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task.
Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity.
The Verification Bottleneck: Why AI's Real Cost Is Human Attention.
Announcing ARC-AGI-3.

The AI Fix
The AI Fix podcast is presented by Mark Stockley.
Grab T-shirts, hoodies, mugs and other goodies in our online store.
Learn more about the podcast at theaifix.show, and follow us on Bluesky at @theaifix.show.
Never miss another episode by following us in your favourite podcast app. It's free!
Like to give us some feedback or sponsor the podcast? Get in touch.
Support the show and gain access to ad-free episodes by becoming a supporter: Join The AI Fix Plus!

Advertising Inquiries: https://redcircle.com/brands

Privacy & Opt-Out: https://redcircle.com/privacy

In episode 97 of The AI Fix, Mark is joined by Ethan Ding, one of the sharpest contrarian voices in AI, to explore why almost half of America's announced data centres aren't being built, and why AI job losses are not what they seem.
Also in this episode, California approves a startup to let AI prescribe antidepressants; Sam Altman pitches a New Deal for the age of superintelligence while sitting on the biggest pile of private capital ever raised; and Anthropic decides it's built a model so dangerous it can't let anyone use it.
Episode links:
Almost Half of US Data Centers That Were Supposed to Open This Year Slated to Be Canceled or Delayed.
AI Psychiatry Startup Approved to Prescribe Meds.
Behind the Curtain: Sam's superintelligence New Deal.
Project Glasswing: Securing critical software for the AI era.
TextQL - AI Data Analytics.
mandates | Ethan Ding | Substack.

The AI Fix
The AI Fix podcast is presented by Mark Stockley.
Grab T-shirts, hoodies, mugs and other goodies in our online store.
Learn more about the podcast at theaifix.show, and follow us on Bluesky at @theaifix.show.
Never miss another episode by following us in your favourite podcast app. It's free!
Like to give us some feedback or sponsor the podcast? Get in touch.
Support the show and gain access to ad-free episodes by becoming a supporter: Join The AI Fix Plus!

Advertising Inquiries: https://redcircle.com/brands

Privacy & Opt-Out: https://redcircle.com/privacy

In episode 96 of The AI Fix, Mark uncovers new research that shows AIs will lie and scheme to save each other, and Mike Fey, CEO of episode sponsor Island, drops in to talk about the future of enterprise work.
Also in this episode, Claude learns to make skills that make skills that make skills; Gemini wants to rummage through your digital life; OpenAI quietly turns ChatGPT into a $100 million advertising juggernaut; and a Waymo robotaxi takes a wrong turn through a school zone.
Episode links:
Skillception — On the Recursive Limits of Meta-Skill Generation.
Google rolling out Personal Intelligence to free Gemini, Chrome, & AI Mode users.
ChatGPT ads hit $100M in six weeks — and OpenAI is just getting started.
Waymo car goes wrong way in Alamo Heights school zone, concerns parents.
AI models will secretly scheme to protect other AI models from being shut down, researchers find.
Island.

The AI Fix
The AI Fix podcast is presented by Mark Stockley.
Grab T-shirts, hoodies, mugs and other goodies in our online store.
Learn more about the podcast at theaifix.show, and follow us on Bluesky at @theaifix.show.
Never miss another episode by following us in your favourite podcast app. It's free!
Like to give us some feedback or sponsor the podcast? Get in touch.
Support the show and gain access to ad-free episodes by becoming a supporter: Join The AI Fix Plus!

Advertising Inquiries: https://redcircle.com/brands

Privacy & Opt-Out: https://redcircle.com/privacy

In episode 95 of The AI Fix, Mark is joined by Dr. Tamara Nall to explore the rise of human–AI relationships and why an estimated 2% of adults in the US are now romantically involved with AI.
They dive into what this means for AI rights, consciousness, and a world where our closest relationships may no longer be fully human.
Also in this episode: human brain cells grown on a chip learn to play Doom; Anthropic accidentally leaks the code behind one of the most talked-about AI tools; ChatGPT isn't getting an erotic mode; and OpenAI's Sora shutdown pulls the rug from a $1B Disney deal.
Episode links:
- OpenAI abandons yet another side quest: ChatGPT's erotic mode.
- How human neurons on a chip learned to play Doom.
- Sora Was Reportedly Costing OpenAI $1 Million Per Day.
- Anthropic goes nude, exposes Claude Code source by accident.
- ReliAI — Your AI Life Operating System.
- The Leading Niche - Data Analytics & Federal Consulting.
- HUMANAI - Sovereign Virtual Space for Digital Citizenship.

The AI Fix
The AI Fix podcast is presented by Mark Stockley.
Grab T-shirts, hoodies, mugs and other goodies in our online store.
Learn more about the podcast at theaifix.show, and follow us on Bluesky at @theaifix.show.
Never miss another episode by following us in your favourite podcast app. It's free!
Like to give us some feedback or sponsor the podcast? Get in touch.
Support the show and gain access to ad-free episodes by becoming a supporter: Join The AI Fix Plus!

Advertising Inquiries: https://redcircle.com/brands

Privacy & Opt-Out: https://redcircle.com/privacy

In episode 94 of The AI Fix, Mark digs into what really scares people about AI and is surprised to learn it isn't robot dogs with flamethrowers, while guest host Alfred Nutile shares his three steps to being less terrified of AI.
Also in this episode, Mark has a bath with his Mac; Yann LeCun gets a billion dollars to prove the entire AI industry wrong; Atlassian fires 1,600 people to fund its AI pivot; and somebody finds a way to make money from Spotify, and then promptly gets arrested for it.
Episode links:
DailyAI Studio.
Put Claude to work on your computer.
Yann LeCun just raised $1bn to prove the AI industry has got it wrong.
Atlassian follows Block's footsteps and cuts staff in the name of AI.
North Carolina Man Pleads Guilty To Music Streaming Fraud Aided By Artificial Intelligence.
What 81,000 people want from AI.

The AI Fix
The AI Fix podcast is presented by Mark Stockley.
Grab T-shirts, hoodies, mugs and other goodies in our online store.
Learn more about the podcast at theaifix.show, and follow us on Bluesky at @theaifix.show.
Never miss another episode by following us in your favourite podcast app. It's free!
Like to give us some feedback or sponsor the podcast? Get in touch.
Support the show and gain access to ad-free episodes by becoming a supporter: Join The AI Fix Plus!

Advertising Inquiries: https://redcircle.com/brands

Privacy & Opt-Out: https://redcircle.com/privacy