We are entering a new epoch where the ability to use and to work with artificial and synthetic intelligence is a human right. Access to intelligence – the prerogative to innovate, work with, and benefit from higher levels of synthetic intelligence – belongs to the people.
Building on increasingly inexpensive compute, abundant data, and low-cost, open-source models, we are about to witness a synthetic intelligence cornucopia.
To support this evolution, we must build infrastructure that fosters pluralistic development of AI. This is the motivation behind The Thames Network, based at Oxford: a decentralized intelligent network designed to operate at the edge, enabling private, censorship-resistant, depoliticized, and decentralized AI facilitated by economic incentives and cryptographic proofs. Principal scientist, Oxford Professor Philip Torr, emphasizes the importance of addressing the concentration of power and the loss of privacy that AI may bring. He advocates for robust technical solutions, such as blockchain, promoting continuous progress in decentralized AI that empowers the masses.
Decentralizing AI
What does this open-source, decentralized marketplace, protocol, and incentive layer for artificial and synthetic intelligence entail? The launch of DeepSeek signifies a shift; open-source AI is not just a trend but a pillar of the future, challenging the hegemony of large centralized corporate or state-run models. AI is migrating from a centralized hub to a decentralized edge.
Recent announcements from Microsoft reveal that distilled, NPU-optimized versions of DeepSeek R1 will soon be available on PCs. These innovations leverage on-device, local processing capabilities, starting with Qualcomm Snapdragon X and extending to Intel Core Ultra 200V and other platforms. This enables users to engage with cutting-edge models entirely locally.
Neural Processing Units (NPUs) serve as microprocessors designed to emulate the brain’s neural function, providing efficient inference capabilities. This unlocks a paradigm where generative AI doesn’t just respond to direct commands but can facilitate semi and fully continuous operations—essentially, it optimizes the role of agents in real-time.
This migration towards decentralization goes beyond mere technological upgrades. It reflects a profound shift in how we choose to empower individuals at the grassroots level, promoting AI systems that encourage collaboration while safeguarding against the pitfalls of centralized control. Renowned AI pioneer Richard Sutton has pointed out that reinforcing learning—rather than large language models—holds the key to driving the future of AI.
Democratizing AI
The Thames Network democratizes access to AI through the development of the first open-source decentralized AI marketplace, protocol, and incentive layer.
While some AI oligarchs, including Elon Musk, advocate for Universal Basic Income as an answer to the job displacement caused by automation, this approach often emphasizes corporate needs over individual empowerment. Instead, we propose a model that democratizes access to AI, enhancing personal autonomy and sovereignty.
With decentralized intelligence at the edge and a revolutionary economic framework, a new intelligent network could emerge, illuminating pathways for collaborative human-AI interactions rather than replacing human roles.
Depoliticizing AI
For artificial and synthetic intelligence to genuinely serve humanity, it must operate independently of biases and political agendas. Censorship, restrictions, and access limitations based on unyielding criteria such as jurisdiction or cost hinder our collective ability to engage effectively with AI.
Privacy considerations remain paramount—especially in sensitive areas like healthcare. Therefore, a decentralized intelligent network must incorporate a privacy-first design, built upon a framework of trust that emphasizes a zero-trust security model while navigating governance and compliance challenges.
The initial phases with hundreds of thousands of models are expected to escalate into an expansive wave of hundreds of millions of domain-specific models, enhanced through Retrieval-Augmented Generation (RAG). The Thames Network aims to provide the necessary tools and an open marketplace for individuals to construct and monetize their specialized knowledge, reinforcing the concept of human-AI partnership.
Reflecting on the ongoing digital transformation, Bill Roscoe, Director of the Oxford Blockchain Research Centre, asserts, “The world really needs an altruistic development of the rules of digital civilization and an infrastructure to support and govern it in a truly collective way.” The Thames Network’s mission is to prioritize privacy and collective governance as we advance technologically.
The convergence of decentralized computing, blockchain solutions, governance frameworks, crypto incentives, and expertly curated AI models sets the stage for a future defined by transparency, accessibility, security, and collaboration. Through The Thames Network, announced during the Oxford AI x Blockchain conference, we envision a world where both humans and AI thrive synergistically. Anything less would represent a disservice to our roles as technologists, engineers, researchers, and economists.
