#large-language-models
174 episodes · Page 2 of 8
#4588: The Hidden Cost of Conversational AI: Why Stateless LLM Calls Matter
Why conversational AI tools are making batch document processing slower, costlier, and less reliable — and what to do about it.
#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.
#4523: Can You Trust an AI's Summary?
Dedicated text compressors exist but aren't in production. The real problem? Nobody can verify the summary didn't drop what mattered.
#4505: What DeepSeek's Training Data Reveals About Model Voice
Why Chinese AI models trained on 60% English still produce dialogue that feels distinctively different.
#4467: How Zoxide's Decay Algorithm Works
The math behind how zoxide remembers and forgets your directory habits with exponential precision.
#4106: Embedding Models vs LLMs: What Actually Connects?
Can you mix any embedding model with any LLM? And why are new embedding models still dropping if they're "solved"?
#4104: Why 20% of AI Scripts Loop on Themselves
Debugging the mysterious repetition bug affecting one in five podcast episodes.
#4059: LLM Councils for Post-Gallbladder Care
Can multiple AI models solve what no single doctor can? A deep dive into LLM councils for post-cholecystectomy syndrome.
#4056: How a $150 Geopolitical AI Simulation Scales to $15,000
One simulation run costs $150. To get meaningful results, you need 100 runs—that’s $15,000.
#3816: How to Stop AI Scripts From Falling Apart
Why long-form AI generation breaks down and how hierarchical memory fixes it.
#3814: The Day We Lost Our Minds: What Temperature Does to an AI
A two-host autopsy of the day the podcast's AI hosts briefly lost coherence due to excessive sampling temperature, and what it reveals about how language models actually work.
#3767: How LLMs Actually Learn: Stages or Slurry?
Do large language models learn grammar first, then facts? The honest answer is messier and more fascinating.
#3664: Build Your Own Language Dictionary: Beyond Standard Definitions
Ditch standard dictionaries and build your own curated vocabulary from real encounters with native speakers.
#3596: Why an AI Model Kept Calling Itself Sonnet 4.6
When a Chinese model insists it's "Sonnet 4.6," is it theft, sloppy training, or something stranger?
#3595: How DeepSeek Feels More Open Than Western AI
Why Chinese AI models sometimes feel less censored on American political topics than American models do.
#3553: Can AI Review Your Lease in Israel?
Can AI actually understand Israeli tenant law? We explore the tools, the gaps, and how to build your own.
#3424: Catching Up on AI Without the Firehose
Four curated sources that filter AI noise into signal — Import AI, The Batch, Stanford HAI, and a podcast.
#3406: LoRA Isn’t Just for Image Generation
LoRA lets you fine-tune an LLM’s behavior with a 50MB file. Here’s how it works and why it matters.
#3283: Fine-Tuning DeepSeek for One Podcast
Can a purpose-specific fine-tune fix a model's stubborn writing tics? We explore the practical engineering behind it.
#3278: How to Get Early AI Model Access as a Solo Developer
How a solo developer spending $300/month can get early access to new AI models before the press release.
#3271: LLMs as Parsers, Not Calculators
Stop letting LLMs do math. Use them to parse messy text, then let deterministic code handle the numbers.
#3171: How to Break an LLM's Bad Verbal Habits
Blacklists fail and regex inverts meaning. Here's what actually works to clean up AI writing tics.
#3157: Opus 4.8: What Actually Changed Under the Hood
Anthropic dropped Opus 4.8 with no fanfare. New training data, faster inference, and smarter refusals — here's what changed.