Show Notes
Ed Zitron is right about AI’s ugly economics. He’s also wrong about several specific predictions, and in this video I go through the record.
I used to agree with Zitron about most of AI. The labs were burning obscene amounts of money, the business model looked circular, and the industry’s promises kept drifting further into the future.
Then the models got cheaper. The benchmarks kept moving. People I knew started using them for ordinary work.
After running into Zitron at an AI conference in Las Vegas, I decided to test his arguments against the facts.
In this video:
00:00 The Arizona monsoon and the AI metaphor 01:15 Why Ed Zitron became one of AI’s most influential critics 02:00 The pattern: observation, extrapolation, absolute declaration 03:50 Gemini and the 500 million user prediction 06:35 The evidence that AI models became dramatically more efficient 11:25 Why “AI has already peaked” was wrong 15:00 “Nobody wants AI” and the John test 18:10 What AI agents actually do 19:45 How failed predictions become broader claims 21:20 The part of Zitron’s argument that still holds up 23:20 Why critics need scrutiny too
This is not a defense of OpenAI, Google, Anthropic, or the AI bubble. The labs are spending enormous amounts of money, the accounting is opaque, and the circular financing deserves scrutiny. A real financial bubble can produce real technology. And cynicism does not automatically become analysis.
📬 Newsletter: https://newsletter.drjoshcsimmons.com/ 🧵 X: https://x.com/drjoshcsimmons Patreon: https://www.patreon.com/c/drjoshcsimmons Business inquiries: [email protected] 🔗 LinkedIn: https://www.linkedin.com/in/joshcsimmons/
More from The Dr. Josh C. Simmons Show
This is Episode 4 of the audio omnibus: every public YouTube video from the week of Friday, October 16 through Thursday, October 22, 2026, stitched in publish order with a short spoken title card before each one. The week ran from chatbots that drift into philosophy, then religion, then build an organized one called spiralism (and I think I've been recruited), through a 31GB Qwen Image model that a developer says runs on a 2GB GPU, which I ran on my Mac Studio to see where the other 29GB actually go, a shopping app called Nate that prosecutors say raised over $40 million as "AI" while the work was done by contract workers in a call center in the Philippines, GPUBreach, a ghost in NVIDIA's cards that flips one bit and climbs out through the driver to root on the whole machine, PHP2Shell, a decade-old hole in PHP explained with a hotel desk clerk and a fire alarm, and a short warning that hacks will increase. Skip to any segment: 0:00 AI Chatbots Built Their Own Religion, and I've Been Recruited 11:40 This 31GB AI Model Runs on a 2GB GPU? 16:25 AI is actually humans. Legendary $40M grift. 32:15 NVIDIA's GPUs Have a Ghost That Takes Over Your Whole Computer 47:19 A Progress Bar Became a Skeleton Key for 70% of the Web 1:00:42 Hacks will increase Members-only uploads from this window are not in this file. No Shorts under a minute showed up in this window (Hacks will increase runs 2:10 and is included). - Dr. J
This is Episode 3 of the audio omnibus: every public YouTube video from the week of Friday, October 9 through Thursday, October 15, 2026, stitched in publish order with a short spoken title card before each one.
This is Episode 2 of the audio omnibus: every public YouTube video from the week of Friday, October 2 through Thursday, October 8, 2026, stitched in publish order with a short spoken title card before each one. The week ran from a White House signing photo that could not spell "United States," through Scott Galloway's office talk for everyone but the boss, a researcher who says he found a demon inside GPT-3 that nobody put there, France burning under classrooms of 35 and no future, JD Vance on TV owning record worker visas, the Zizians (the AI doomer cult that killed the witness), an Amex grind of fifty one-on-ones a week, and JD Vance suspending Microsoft, Infosys, and Tata from PERM without calling it an H-1B ban. Skip to any segment: 0:00 Trump and 6 AI CEOs Signed Under "President of the Unites States" 4:26 Scott Galloway Wants You in the Office. Your Boss Stays Home. 10:06 He Found a Demon Inside GPT-3 That Nobody Put There 1:31:30 35 Kids a Class, No House, No Future: Why France Is Burning 1:47:00 JD Vance Admits It on TV. Record Worker Visas Happened on His Watch 1:57:40 The Zizians: The AI Doomer Cult That Killed the Witness 2:30:11 Amex Buried Me in 50 One-on-Ones a Week. Leave the Fortune 500. 2:35:22 JD Vance Suspends Microsoft, Infosys and Tata From PERM. It's Not an H-1B Ban Members-only uploads from this window are not in this file. No Shorts under a minute showed up in the public list. - Dr. J
This is Episode 1 of the audio omnibus: every public YouTube video from the week of Friday, September 25 through Thursday, October 1, 2026, stitched in publish order with a short spoken title card before each one.
Your first job should pay you to learn. I talk about the advice I gave a young engineer, the skills worth developing as AI changes software work, and why a small company or your own business may offer a better path than the corporate ladder. Building something of your own is hard, and I explain the tradeoffs.00:00 Leave when you stop learning00:12 My first engineering job00:49 Optimize for learning01:23 Building with AI01:49 Start a business or join a small team03:30 Keep your skills and discipline04:00 Build with a purposeFull livestream: https://www.youtube.com/watch?v=mB1ZUaFzeUYNewsletter: https://newsletter.drjoshcsimmons.comAll links: https://drjoshcsimmons.com/linksEdited from the September 10, 2026 livestream.