Hallucinations
306 episodes · Page 4 of 13
#5384: Android ASR Runtimes: LiteRT, ExecuTorch, and Why Your Phone Has No VRAM
Why does your phone have no VRAM number? A tour of Android's runtime layer and what it takes to run ASR locally.
#5274: A Broken Dremel and the Hidden World of Parts Distribution
A dead rotary tool leads down two rabbit holes: warranty economics and how components actually move around the world.
#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.
#5184: AI Is a Number Factory, Not a Wordsmith
Under the prose, every AI model is just matrices of floating-point math. So where does the randomness actually come from?
#5182: DeepSeek V4.1 Flash: 1M Context, 437x Smaller KV Cache
DeepSeek V4.1 Flash landed with a 1M-token window and a KV cache 437x smaller than V1. Here's what actually changed — and why the middle of your co...
#5165: Weighting Memory in a RAG Pipeline
Daniel's memory leg won't fire. Turns out system prompts have no inherent vector-space privilege — but retrieval does have a real knob.
#5118: Model Identity Fraud: Distillation or Data Contamination?
When AI models lie about who they are, is it stolen knowledge or just contaminated training data?
#5113: Can Rival Labs Poison AI Training Data?
A few hundred crafted documents can shift a model's stance. Here's how data poisoning actually works.
#5014: PWA vs Native: One Developer's Home Inventory Dilemma
PWAs cover 80% of use cases. What's the cost of the last 20%?
#4999: Why AI Can't Just Put a Circle on Your Photo
Why multimodal AI models can't simply circle a Dremel bit without rebuilding your entire image from scratch.
#4991: Self-Healing Codebases: Can AI Fix Bugs Autonomously?
How far are we from AI systems that fix their own bugs without human intervention? The answer might surprise you.
#4971: Fine-Tune vs Prompt: Taming Hallucinated Deadlines
Big models add fake deadlines. Small models miss tasks. Which wrongness can you build a system around?
#4958: Multi-Model Transcription with LLM Reconciliation
Using multiple transcription engines and an LLM judge to catch errors in voice-to-text — especially tricky technical terms.
#4855: What Turbulence Actually Does to a Plane
How climate change is increasing clear-air turbulence — and why your plane can handle far more than you feel.
#4813: The $70 Mouse That Changed Everything
How a $70 ergonomic mouse purchase reveals the quiet way we learn to trust AI.
#4809: Can a Podcast Train Itself?
Can a podcast become a self-developing project? We explore RAG vs. fine-tuning for lore memory.
#4784: Host vs Router: The Real USB Bonding Trade-Off
Can you skip the travel router and bond cellular + Wi-Fi directly on your Linux laptop? We test the hardware reality.
#4775: The Evaluator Role No One's Building For
Benchmarks like MMLU are broken. A new role is emerging: the domain-specific AI evaluator.
#4727: Surviving the Negev: Matching VHB Tape to Desert UV & Heat
From Jerusalem UV to freezing nights — which 3M VHB tape actually survives the desert? Part numbers and specs inside.
#4692: Why Printed Colors Never Match Your Screen
Screen light vs. ink absorption: why RGB and CMYK clash, and how ICC profiles, rich black, and bleed save your prints.
#4639: Concrete Walls vs. Wi-Fi: Choosing a Fill-in AP
One AP used to cover Daniel's apartment. Then he moved. Concrete walls changed everything.
#4581: Text In, Text Out: Designing Single-Purpose AI Utilities
Why small, narrowly scoped AI tools need system prompts that say "shut up and return the output" — and how to write them.
#4570: One Negative Prompt Beat a Dozen Positives — Here's Why
Why one blunt "don't" outperformed a dozen careful "do's" — and how to structure your system prompts for maximum compliance.
#4492: What Makes a Container Robot-Ready?
The hidden specs that turn a plastic box into something a robot can pick, read, and move at industrial speed.