Why AI Should Become a Commodity

Summary

Massimiliano Genta argues that the next phase of artificial intelligence should expand participation beyond the companies training the largest foundation models. In his view, people and smaller organizations can improve intelligent systems through the environments, tools, workflows, feedback, and accumulated experience surrounding a model, without building a frontier model themselves. As models become available through common interfaces or local deployment, they could be selected according to cost, performance, privacy, and task requirements, although they will not become fully interchangeable. Genta says the main value may increasingly lie in applying intelligence to real problems and preserving what systems learn. He identifies a risk that agents and applications will repeatedly rediscover the same lessons because their experiences remain isolated. Project 507 is exploring an open learning substrate that records relationships among expected outcomes, actions, actual results, and future decisions so experience can be reused across systems and models. The proposed infrastructure would be open source, while still allowing users to control what private learning is shared. Genta acknowledges that this privacy and autonomy problem remains unsolved. He concludes that democratizing AI should mean giving more people the ability to influence and improve intelligent systems, not merely providing access to a small set of centrally owned models.