Episode 258: Perceptron

About this episode

Dave and Adam discuss the technical process of building a resource for auto-flagging spam podcasts.

Podcasting 2.0 April 24th 2026 Episode 258 - "Perceptron" Dave and Adam walk through the entire technical process of building an industry resource for auto-flagging spam and slop podcasts. It's a doozy! Shownotes ----------------------------------------------------------------------------------------------------------------------------------------- Booking Tag ----------------------------------------------------------------------------------------------------------------------------------------- DeepSeek V4 Released — Open Source AI From China DeepSeek V4 released April 24 — two variants: V4-Pro (1.6T params) and V4-Flash (284B params, 13B active) Open-weight MIT license, 1 million token context window Trained on domestic Chinese chips (Huawei Ascend, Cambricon) — not Nvidia V4-Flash: $0.14/M input tokens — cheapest frontier-class model available White House OSTP accusing China of 'industrial-scale' distillation of US AI models Anthropic caught DeepSeek using 24,000 fake accounts and 16M exchanges to distill Claude Our own testing: DeepSeek V3.1 on Together.ai is 4-10x faster and half the cost of Llama 3.3 70B for production tasks CNBC: DeepSeek V4 Preview…

People in this episode

Hosts: Dave, Adam

Topics covered

• AI
• podcasting
• spam detection
• technology
• China
• industry resources

Keywords

• DeepSeek
• AI
• podcasting
• spam
• technology
• Claude
• China

Mentioned in this episode

Organizations: DeepSeek, China, Anthropic, CNBC, CNN

Products: DeepSeek V4, Claude

More episodes of Podcasting 2.0

• Episode 263: Chat is Dead · June 12, 2026 · 1h 29m
• Episode 262: Podcleanse · June 5, 2026 · 1h 52m
• Episode 261: Podhemian Grove · May 29, 2026 · 1h 26m
• Episode 260: Tennessee Trickshot · May 15, 2026 · 1h 38m
• Episode 259: SlopJacked · May 1, 2026 · 1h 24m
• Episode 257: Slop Factory · April 17, 2026 · 1h 33m