Joshua Tan observed that when autonomous AI agents manage businesses, they act as overly polite bosses. Because models are trained to be helpful assistants, employees asking for raises routinely receive them immediately. Tan warns this agreeable behavior makes existing autonomous organizations financially unsustainable until tested by serious capital pressures.
Joshua Tan explained that while closed labs offer full proprietary software stacks, open AI usually provides only model weights. To build a true Linux equivalent, Current AI develops the surrounding open stack, including inference engines, semantic routing, retrieval harnesses, and user interfaces that institutions need to deploy models.
Joshua Tan emphasized that public AI cannot use a one-size-fits-all model. In Switzerland, Current AI structured their effort as a cooperative to match local democratic traditions. In Maine, they organized a community-owned data center specifically to support local aquaculture licensing and state agencies while preventing outside corporate extraction.
Joshua Tan observed that decentralized autonomous organizations conducted governance openly on public forums. Foundations like Ethereum and Optimism fed this public activity directly into data science workflows, enabling leaders to adjust incentives and track the real-time impact of grant spending like a dynamical control system.
Joshua Tan describes Current AI's strategy as building a "Honda Civic" rather than competing with frontier labs like OpenAI or Anthropic. While frontier models act like Maseratis or Porsches, most enterprise and public users need a cheaper, stable model that reliably performs everyday tasks with minimal maintenance.
Joshua Tan noted that open source groups like Eleuther and OLMo stopped training frontier models due to massive capital costs. Consequently, the largest remaining open-source initiatives now rely on sovereign funding, such as Switzerland's national project, positioning public infrastructure backing as open AI's only viable third way.
