PART THREE
Panel Participants
Moderator
Justin Roberti
Panel
- Gaurav Sharma, io.net
- Toufi Saliba, HyperCycle
- Martin Rezny, Cogensy
The discussion explored one of the defining questions facing artificial intelligence: can AI infrastructure remain open, decentralised and human centred as the technology develops at unprecedented speed? While the panellists approached the issue from different perspectives, all agreed that the future of AI should ultimately serve people rather than concentrate power in the hands of a small number of organisations.
Governance, Human Agency and the Future of Decentralised Intelligence
While the opening discussion focused on infrastructure and identity, the final part of the panel turned to perhaps the most fundamental question of all: if decentralised AI succeeds, who governs it, how are decisions made, and what role do humans continue to play in a world increasingly shaped by intelligent machines?
Those questions were addressed by Martin Rezny of Cognesy, whose background in political philosophy and competitive debating offered a markedly different perspective from the engineering approaches presented earlier. Rather than concentrating on processors, networks or cryptography, Rezny focused on communication itself, arguing that society’s greatest challenge may not be building more intelligent machines, but enabling people to deliberate and make better collective decisions.
“I’m a political philosopher by background,” he explained, describing more than two decades spent studying debate, logic and structured reasoning. Those disciplines now underpin Cognesy’s work developing what he described as a communication and deliberation platform capable of allowing even millions of people to participate meaningfully in large scale discussions.
For Rezny, existing approaches to AI governance remain heavily influenced by utilitarian thinking, reducing complex human questions to optimisation problems. While useful in some situations, he argued that such approaches struggle to represent the diversity of human values or the richness of public reasoning.
Not just Utilitarianism
“There is more to politics and more to philosophy and more to reasoning than utilitarianism.”
Instead, Rezny suggested that many of the tools already exist within the humanities. Socratic dialogue, structured debate and centuries of philosophical practice provide tested methods for examining arguments, exposing assumptions and reaching more thoughtful conclusions. Artificial intelligence, he argued, offers an opportunity to scale those methods beyond anything previously possible.
Rather than replacing human judgement, AI could become a facilitator, helping individuals organise arguments, compare competing viewpoints and participate in discussions that would previously have been impossible because of their size or complexity. In that sense, technology would strengthen democratic participation rather than diminish it.
The discussion prompted an immediate response from Tufi Saliba, who questioned whether some of the problems being described were problems at all. He cautioned against framing AI primarily through fear, arguing that many organisations developing advanced systems genuinely believe they are contributing positive solutions rather than creating threats.
“I wasn’t aligned with the way you described the problem,” Saliba told Rezny.
Rezny agreed that his position had perhaps been oversimplified and clarified that his objective was not to criticise technology itself but to improve the quality of human reasoning feeding future AI systems.
“What we want to provide is a better platform for humans to reason, so that it produces better data about human reasoning.”
If humanity is helping to educate increasingly capable AI, he argued, then improving the quality of human deliberation becomes one of the most important contributions society can make. Better discussions produce better reasoning, and better reasoning ultimately produces better intelligent systems.
Throughout the exchange, moderator Justin Roberti repeatedly steered the conversation back towards practical outcomes for ordinary users. While the philosophical discussion ranged from artificial superintelligence to political theory, he challenged the panel to explain what success would actually look like for people using AI in their everyday lives.
“What is the future that we want our users to be looking forward to?”
Rather than abstract debates about technology, Roberti suggested that AI should ultimately be judged by whether it empowers individuals, creates opportunity and remains centred on human needs.
Saliba welcomed that framing.
“I’m a huge proponent of decentralization as a method,” he said, before adding that the discussion should always begin with the objective everyone shares.
Human Centric
“What is one thing that everyone is agreed on? Let’s make it human centric.”
Although the panel contained clear differences of opinion over regulation, governance and implementation, that statement became one of the strongest points of consensus. Each participant approached decentralisation from a different discipline, yet all argued that AI should expand human capability rather than diminish it.
Roberti closed the discussion by asking each speaker how society would know when decentralised AI had truly succeeded.
Rezny’s answer returned to the individual.
“I think it’s when users are able to use AI in their own ways differently across the world for practical purposes that they care about personally.”
Current AI systems, he suggested, remain too general and often insufficiently reliable for many everyday tasks. Success will come when people stop thinking about AI as a technology and simply begin using it naturally to solve problems that matter in their own lives.
As the session drew to a close, the discussion revealed four complementary perspectives on the future of decentralised intelligence.
In Conclusion
Justin Roberti framed the debate by asking whether AI infrastructure can remain open and accessible rather than becoming concentrated within a handful of technology companies.
Guarav Sharma argued that affordable, decentralised compute is essential if innovation is to remain competitive and available to builders beyond the world’s largest corporations.
Tufi Saliba looked further ahead, describing a future in which cryptography, identity and interoperable AI systems provide individuals with greater control over their digital lives while enabling new forms of collaboration between humans and intelligent agents.
Martin Rezny reminded the audience that technology alone is insufficient. Human reasoning, governance and meaningful deliberation remain essential if increasingly powerful AI systems are to reflect the diversity of the societies they serve.
Taken together, the discussion suggested that decentralised AI is no longer simply a technical challenge. It has become an interdisciplinary endeavour, bringing together infrastructure, philosophy, economics, governance and human values. The future will almost certainly require advances in all of those fields if artificial intelligence is to fulfil its promise while remaining firmly centred on the people it is intended to serve.
