#multimodal-ai
30 episodes
#5903: Video Tokens, Omni, and the Local VRAM Math
Video costs ~100 tokens per second. That single number explains why ChatGPT hides video upload and why local inference hits a KV cache wall.
#5626: What Multimodal Models Actually Learn From Video
Late fusion, early fusion, and vision laziness — what actually happens when models train on video, audio, and text together.
#5301: Why Talking Robots Are Really a Committee
That humanoid chatting while it moves? It's not one brain — it's a stack of separate models glued together.
#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.
#4834: Why Your AI Bill Keeps Surprising You
How a 37-cent project and a $100 simulation reveal the hidden math of AI API pricing.
#4774: When a Token Isn't a Word Anymore
Images, audio, and code all use "tokens" — but they're completely different things. Here's how tokenization actually works across modalities.
#4668: Why Podcast AI Voices Sound Too Perfect
We dig into why AI podcast voices sound too clean—and how TTS is learning to stumble, overlap, and interrupt convincingly.
#3234: Who Should Sponsor This Podcast? Open-Book Economics
We open the books on our AI-generated podcast: $200/month costs, 180K plays, zero sponsors. Who should we pitch?
#2688: Intelligent Frame Extraction for Multimodal AI
Use multimodal AI and smart frame extraction to turn a walk-through video into an actionable decluttering plan.
#2409: When AI Cheats on Cultural Knowledge
Five benchmarks that reveal how AI systems fail at cultural knowledge — and what their methodologies tell us.
#1964: The Three Layers That Make AR Finally Work
See a 3D arrow pointing to the exact bolt you need, or read a street sign in real-time translation.
#1792: Google's Native Multimodal Embedding Kills the Fusion Layer
Google’s new embedding model maps text, images, audio, and video into a single vector space—cutting latency by 70%.
#1724: When AI Dubbing Swaps Your Gender
How does YouTube translate a video with one click? We explore the tech behind auto-dubbing, from sandwich models to voice cloning.
#1592: The Vector Debt Trap: Choosing Embeddings That Last
Stop treating embedding models like plumbing. Learn how to navigate vector debt, multimodal retrieval, and database configuration for RAG.
#1586: The Rocketbook Sunset and the Search for a Clean Erase
Bridge the gap between handwritten notes and AI. Discover the best whiteboard notebooks and markers for seamless digital transcription.
#1584: Beyond Text: How Gemini 1.5 Flash Is Revolutionizing Audio
Discover how native multimodality in Gemini 1.5 Flash is killing the "transcription tax" and enabling deep forensic audio analysis.
#1568: The Signal Versus Symbol Gap
Is Gemini a brilliant audio engineer or just a talented lip-reader? Explore the "signal vs. symbol" gap in AI audio processing.
#1564: The Death of the Cascaded Pipeline
Forget basic transcription. Explore how native omni-modal models are capturing the "soul" of speech with near-instant latency.
#1557: Why 95% of FDA-Cleared AI Fails to Help Patients
Explore the shift from simple AI detection to multimodal systems and the growing challenge of deepfake medical images in healthcare.
#1482: The Hidden Cost of Choosing an Embedding Model
From Matryoshka models to multimodal search, discover how the fundamental units of AI memory are being optimized for efficiency and scale.
#1322: When AI Stops Seeing and Starts Reasoning
Explore how AI evolved from simple pixel labeling to understanding intent and context through Vision-Language Models and agentic frameworks.
#1085: The Tokenization Lie: How AI Actually Processes Media
Think 1,000 tokens equals 750 words? For audio and video, that rule is a lie. Discover the hidden math behind multimodal AI.
#786: The Cost of a Touch: When Your Hoard Becomes a Liability
Learn how to manage thousands of parts without losing your mind using AI, QR codes, and professional logistics strategies.
#769: When Manuals Learn to See in 3D
Discover how AI and spatial computing are turning complex hardware repairs into real-time, interactive experiences.