AI Evaluation
516 episodes · Page 6 of 22
#5436: Small Models as Rewriters, Not Writers
Why "don't say X" prompts backfire, and how a tiny grammar-constrained model can scrub a script without breaking its grammar.
#5431: The Other Half of Hugging Face: Why BERT Still Out-Downloads Llama
Encoder models pull over a billion downloads a month. Decoder models pull 397 million. The AI conversation and the download counter are describing ...
#5426: AI Wrote His Yom Kippur Prayer
A man skipped synagogue, opened ChatGPT instead, and got a prayer that knew his life. Is AI-generated prayer real prayer?
#5412: Editing vs. Note-Taking for AI Fine-Tunes
Hand-editing a model's output gives three training signals at once. Writing notes gives one — and a weaker one at that.
#5409: What Makes a Model "Agentic"? Atria Dawn Preview
A 744B-parameter model drops quietly on Hugging Face. Is "agentic" a real model category, or just a deployment pattern?
#5408: When Small NLP Models Beat the LLM
Feature extraction, fill-mask, token classification — the classic NLP tasks still have a job. Here's when a small model beats a frontier API.
#5407: Hemmingway-1 and the War on Waffle
A 27B model promises answers without the preamble. Its benchmark is homegrown — and the behavior it targets has a paper trail.
#5401: When Companies Hide Humans Behind the AI Curtain
A system prompt and a cheap model can make human decisions read like bot output — and that's exactly the point.
#5399: Who's Actually Behind the AI? Fauxtomation Explained
Amazon's Just Walk Out needed 1,000 workers in India. Google Duplex, Facebook M — the pattern of selling humans as software goes deep.
#5391: The Software Behind Urgent Care Triage
Big buttons, emoji vitals, and a system that says "order IV" — what's actually running behind the triage screen?
#5312: Fighting Fake-Personalized Spam With AI Sinkholes
Scraped flattery, fake familiarity, and the one-sided ask — plus the tarpit history behind Daniel's sinkhole agent idea.
#5276: Who Really Invented the Gantt Chart?
Karol Adamiecki built it in 1896 — in Polish. Henry Gantt got the name. And why critical path analysis still catches what humans miss.
#5256: Ruby Won the AI Benchmark. Is It Dying Anyway?
97% of public Ruby repos sat untouched for six months — yet Ruby beat Python in an AI coding benchmark. What's actually declining?
#5255: When AI Writes the Code, Who Checks It?
If agents write the code, the human job becomes knowing when it's wrong. What that means for languages, jobs, and CS degrees.
#5251: Recovering the Moon That's Already in the Pixels
Image restoration clarifies photos without inventing detail — until the fidelity dial shows up. Here's what's in the toolbox.
#5230: Why Your Research Agent Won't Stop Searching
Most agent search runs add no new evidence — and 66-85% of the context window is just search snippets. Here's what actually stops the loop.
#5187: Translation Is Not a Pipe: What Machines Miss
Rubrics rate machine translation above human work — until readers pick side-by-side. What translators know that metrics can't see.
#5183: Where AI Actually Sits on the Hype Cycle
AI has been doing real work for decades. So why do vendors still shout it from the rooftops?
#5169: Fine-Tuning vs From-Scratch for Minor Languages
What 774 experiments and a new Armenian model reveal about the tradeoff between fluency and knowledge in low-resource languages.
#5167: The Stethoscope Isn't Dead, It's Getting AI Ears
Two centuries of the stethoscope — and why AI is now making it more useful than ever, not obsolete.
#5154: DeepSeek's Two-Endpoint Philosophy
DeepSeek quietly routed Pro traffic to Flash — and that routing change says everything about its two-endpoint strategy.
#5097: Who Actually Gets Paid to Test Hardware?
Real testing vs. affiliate content — and whether AI can finally separate genuine reviews from the noise.
#5088: What Does a Pilot's Gray Hair Actually Buy You?
Is flight seniority about skill or something harder to measure? We dig into what experience really buys in the cockpit.
#5056: How AI Writes a 30-Minute Podcast in One Pass
What does it take for a model to write a full episode without edits? The answer involves context windows, repetition, and a surprising architecture...