We Study Billionaires: Infinite Tech – "AI for Activists"

Guests: Justin Moon & Srumenik

Host: Preston Pysh
Date: September 17, 2025

Episode Overview

This special edition of Infinite Tech explores the intersection of artificial intelligence, activism, and decentralization with esteemed guests Justin Moon and Srumenik—leading engineers in the Bitcoin space now pioneering in AI. The episode dives into enabling activists with new AI tools, privacy concerns, the balance between centralized and open models, and the flourishing world of decentralized, open-source tech.

Key Themes & Discussion Points

1. The New Era of Coding: Vibe Coding & AI as a Tool

• Vibe Coding Demo with Jack Dorsey

  • Justin shares an anecdote about live coding with Jack Dorsey at the Oslo Freedom Forum, where they collaboratively built a regulatory suggestion site using Jack’s open-source coding agent—demonstrating the power and spontaneity of modern AI-assisted coding.
  • "Vibe coding is...non-deterministic. You might get something good, you might not." – Justin (04:03)

• Empowering Non-Programmers

  • AI tools lower the barriers to entry for software creation, removing the “semicolon jail” of traditional coding and enabling a broader set of users.
  • "Now you can utilize these AI tools…for a lot of people, you don't have to worry where all the little semicolons go..." – Justin (05:11)

• AI as a Personal and Business Productivity Lever

  • Jack Dorsey's daily practice of dedicating three hours to exploring AI coding underscores how mainstream this has become, even for CEOs.
  • "He’s always trying to push the limit of what's possible." – Justin (05:39)

• Capital Costs and AI Accessibility

  • AI is making software development more capital-intensive—the best results now come with higher compute spend.
  • "It’s turning software into a bit of a capital intensive endeavor... What if it's $2,000 a month?" – Justin (09:46)

2. AI for Activists: Human Rights Foundation Initiatives

• HRF’s AI for Individual Rights Program

  • HRF recognized AI's value to activists, especially for accelerating time-consuming tasks like grant writing, and launched educational/workshop initiatives.
  • "All these activists found ways to write a grant four times faster, right, using ChatGPT." – Justin (16:58)

• Grassroots Tools: Bitchat & Cashew

  • Open-source, privacy-centric protocols (e.g., Bitchat—Bluetooth mesh messaging, and Cashew—offline bitcoin transactions) enable secure communications and exchanges even in censored or blackout environments.
  • "It's like a backstop of freedom, something that can't be taken away unless they take your phone." – Justin (21:47)

• Decentralized Application Development

  • Tools like Bitchat are rapidly iterated—sometimes in a week—and quickly ported across platforms using AI to “translate” code, spawning robust community-driven ecosystems.
  • "He was able to get it to translate it from one language to another...in about a week." – Justin (19:47)

3. Decentralization, Privacy, and Open Source AI

• Challenges and Opportunities of Decentralization

  • Training large models remains capital intensive and favors centralization (big GPU farms), whereas inference (model use) is increasingly decentralized.
  • "For training the best models...you need the most amount of GPUs to the smallest place possible." – Srumenik (38:46)

• Synthetic Data and Model Proliferation

  • Teams like DeepSeek have made huge leaps by generating synthetic data from ChatGPT, allowing quickly trained, lower-cost, and competitive models.
  • "They created a bunch of synthetic data using ChatGPT...so they extracted knowledge out of ChatGPT to train it." – Srumenik (34:06)

• AI’s Democratization Progress

  • The landscape has shifted from a few centralized LLMs to a robust, competitive ecosystem across continents, thanks to data accessibility and reverse engineering.
  • "A year ago…there were very few participants. Now it's very unclear who has the best AI." – Justin (32:16)

4. The Future: Specialized Models, Robotics, and Human Potential

• Cost Curve and Model Scalability

  • As AI models grow, costs could become prohibitive, shifting value toward more specialized, domain-specific models rather than all-encompassing LLMs.
  • "It's getting to the point where it's cost prohibitive…another generation or two and it's going to be cost prohibitive to run thing on the frontier..." – Justin (41:57)

• Comparing AI & Human Intelligence

  • The discussion draws parallels between neural resources in the human brain (e.g., motor cortex versus language) and the challenge of embodying AI for robotics.
  • "The person has embodied and has experience and AI doesn't..." – Justin (44:56)
  • "We don't only reason in language…maybe some reasoning is happening in the spatial part of your brain." – Srumenik (46:19)

5. Education, Customization, and Work Transformation

• AI as a Force in Education

  • Anticipation for personalized education—AI-powered systems tailored to individual interests, aptitudes, and removing instructor bias.
  • "I'm excited about…customization to a person's natural interests and talents…you're going to have the best instruction ever." – Preston (50:04)
  • "An AI could be much better at…being opportunistic than a textbook." – Justin (51:22)

• Automation, Creativity, and Quality of Life

  • The future may see reduced screen time, increased creative agency, and more people contributing to open-source and activist projects.
  • "The skill to do it just kind of went down by maybe an order of magnitude." – Justin (26:34)

Notable Quotes & Memorable Moments

• On AI Blending with Activism:
  "It's a backstop of freedom…unless they take your phone." – Justin (21:47)

• On a New Wave of Programmers:
  "The skill to do it just went down by…an order of magnitude." – Justin (26:34)

• On Decentralization Progress:
  "Now it's very unclear who has the best AI…It's a lot better to have three options than one." – Justin (32:16)

• On Open Source Resilience:
  "If the creators blow it, someone else will carry the torch…All you need is the skill to do it, and that's where AI comes in." – Justin (26:34)

• On AI in Education:
  "An AI could be much better at... being opportunistic than a textbook." – Justin (51:22) "You're going to have the best instruction ever…no ego, no past experience of the teacher themselves…" – Preston (50:04)

Timestamps for Core Segments

• Introduction & Vibe Coding (00:00–06:47)
• AI for Activists & HRF Initiatives (16:05–22:48)
• Open Source Tools: Bitchat & Cashew (18:37–26:34)
• Decentralizing AI: Challenges & Synthetic Data (31:28–38:46)
• Model Training vs. Inference (39:27–41:57)
• Specialization and the Future of Large Models (41:57–44:56)
• Human vs. AI Intelligence & Embodiment (44:56–46:46)
• Education and Individualized Learning (47:39–52:53)

Episode Takeaways

• AI is empowering both programmers and non-programmers, especially activists, democratizing tools previously held by technical elites.
• Rapid iteration and open-source ecosystems are redefining what’s possible in both communication and censorship resistance.
• Decentralization is making strides, especially in model usage, despite centralized bottlenecks in model training.
• Education and work are on the brink of profound transformation—tailoring learning and enabling broader creativity.

Resources & Further Info

• [Justin on Nostr]
• [Srumenik on X/Nostr]
• [HRF AI for Individual Rights Program]
• [Rodstone Project]

(See show notes for full link list)