HeyJaredAI Diatribe
Breaking down AI—one rant at a time! Real talk on AI’s biggest breakthroughs, boldest claims, and real-world impact. AI Diatribe delivers sharp insights that cut through the noise. Follow for bold takes, fresh perspectives, and the occasional AI rant.
- Indexed episodes, last 90 days
- 5
- Latest publication
- Aug 27, 2026
- Audience
- Checking…
- Earliest in this view
- Jul 9, 2026
Latest episodes
Episode 47: We Hold AI Science To A Standard Humans Never Meet (opens the original)
Read excerpt
ChatGPT proposed adding an extra oxygen source to a reaction that already pulls oxygen from air. PhD chemists looked at it and said no, that's not how it's done, you'll spawn side reactions and make a mess. The team ran it anyway. Yield went from 16.6% to 25.2%. Piotr Byrski, co-founder of Molecule.one, built his lab in 2022 with AI as the intended user, before most people had heard the term agentic AI. The number worth holding onto isn't the discovery. It's the volume behind it: roughly 10,000
Episode 46: The Bottom Rung Of The Career Ladder Is Gone (opens the original)
Read excerpt
A Berkeley computer science course went from roughly ten percent of students failing to nearly forty-five percent, and the students found out on the last day before finals ended. That's the enforcement approach. The other approach, same campus, same semester: a writing professor who told her class she knew they were using AI and had them write the usage rules together. Five students scattered across Japan, the Philippines, Malaysia, and the US - education, public health, finance, business law, E
Episode 45: With AI, You're Either at the Table or on the Menu (opens the original)
Read excerpt
A human who blows one deadline loses your trust for good. An AI blows them constantly, and we shrug, tweak the prompt, and try again. Wes Durow has a theory about why, and about what that double standard is quietly costing marketing. Wes teaches marketing at UT Dallas, ran marketing as a CMO, and spent three decades in tech. He treats "AI slop" as a strategy failure, not a tooling one. He explains why your model starts agreeing with you too much, and how that makes your work worse, not easier. A
Episode 44: AI Never Needed Skynet, the Algorithms Already Took Us (opens the original)
Read excerpt
Matt flips the show and interviews Jason, and the first honest thing on the table is that neither of them fits the camps everyone else has picked. One scores his P-Doom at a flat five out of ten and still writes like a pessimist. The other admits he's biased and simply doesn't show it. Then it gets specific. Matt bought a car with Claude in his pocket. The model talked him out of the first purchase, priced out fuel and insurance over five years, found a better car, then coached him line by line
Episode 43: Why Doesn't More AI Training Close the Skills Gap? (opens the original)
Read excerpt
Senior leaders keep repeating AI buzzwords they can't actually define. Dr. Mechie Nkengla has watched it happen for two decades of consulting, and she's stopped being polite about it. Mechie breaks AI literacy into four pieces: responsible AI, risk awareness, human judgment, and operational readiness. She also explains why the EU AI Act now requires companies to prove, through audit, that training changed behavior, not just that people sat through a course. The conversation gets sharper from the
Publishing over time
Last 90 days. Choose a month to open its work.
Recurring subjects
Named in the text we hold. One piece can cover several.
Audience
No verified audience measurement yet.
About this data
Counts cover the work we have indexed. Tone needs enough text and a confident classification. Excerpts and episode notes are not full articles or transcripts.
Identity or attribution wrong? Suggest a correction.