Episode Description: Is spending millions to fine-tune an AI model a waste of resources when frontier models like Gemini and GPT-5 are evolving so fast?

In the debut episode of Distributed Dissent, En Hong (CEO, Generis AI) and Mathias Bock (CEO, Tokuma Labs) discuss the reality of being lawyers-turned-founders in Hong Kong and Tokyo. They debate the diminishing returns of training custom models, the psychological traps peoplefall into when using AI, and the stark differences between building startups in different Asian markets.

Topics Covered:

• Frontier Models vs. Fine-Tuning: Why even well-funded startups are giving up on fine-tuning their own models in favor of riding the wave of big tech scaling.
• The "Gell-Mann Amnesia" Effect: Why experts immediately spot hallucinations in their own field but blindly trust AI in areas they don’t understand—and the risks this poses for legal professionals.
• Hong Kong vs. Tokyo: A comparative look at the startup ecosystems, including the surprising arbitrage in engineering talent and the cultural differences in venture capital.
• Involution hits the Legal Field: How extreme price competition in the region is driving fees down, and what this signals for the future of "median" legal work versus true expertise.
• Manufactured Intelligence: The economic implications of intelligence becoming a cheap utility, and why "plumbers might be safer than partners."

Mentioned in this episode:

• Book: Permutation City by Greg Egan
• Book: Blindsight & Echopraxia by Peter Watts
• Article: Why Birds Don't Drive Bentleys and Why Humans Will Never Fly
• Case Law: Getty Images vs. Stability AI (UK Ruling)