Can we decentralize artificial intelligence?

Artificial intelligence is growing at an extraordinary speed. Large companies are building huge data centers, buying GPUs and accumulating increasing amounts of data. But there is a question that we rarely ask: Who controls all that knowledge? In this episode we explore an idea: what if we could decentralize not only computing power, but also the data, training and knowledge generated by artificial intelligence? We start from simple concepts—data, training, GPUs, blockchain and hashes—until we reach a more ambitious proposal: a network where researchers, universities, developers and users can contribute data, computational capacity and training, receive incentives and allow others to reuse that knowledge. But we also put the idea to the test. Is this really new? Aren't there already decentralized GPU marketplaces? What difference would there be between a network like IAChain and projects like Akash, Bittensor or Gensyn? Is it possible to verify a training? How is private information protected? Can a decentralized network compete with tech giants? This episode is not intended to claim that all of these questions are resolved. On the contrary: it seeks to open the discussion. If Bitcoin raised the possibility of decentralizing money, can we also decentralize artificial intelligence?