Unlocking the Future: From Nexus’ Warnings to Decentralised AI’s Promise

Yuval Noah Harari, in his thought-provoking book Nexus: A Brief History of Information Networks, paints a sobering picture of the challenges posed by emerging technologies like artificial intelligence (AI). As a historian, Harari’s vantage point provides unparalleled insight into how power dynamics, societal structures, and human values can be disrupted by new tools. However, his critique often leans heavily towards the risks and presents a warning of AI’s potential to create “multipolar traps” where competition among actors spirals into unintended consequences. This perspective, while important, may lack the technical lens to envision how decentralised technologies can offer solutions—not traps, but what we might call “multipolar wins.”

The convergence of AI and decentralised technologies offers a transformative roadmap. By harnessing distributed ledger technology (DLT), Zero Knowledge Proofs (ZKPs), Decentralized Autonomous Organizations (DAOs), Fully Homomorphic Encryption (FHE), and other cryptographic innovations, we can construct a future where AI works for communities, rather than concentrating power. Let’s explore how decentralisation, far from being an abstract ideal, is the key to building trustless, scalable, and community-driven AI systems.

The Problem Nexus Highlights: A Multipolar Trap

Harari’s concern revolves around the emergence of hyper-competitive AI ecosystems where stakeholders act in self-interest, leading to:

1. Data Monopolies: Centralised AI systems consolidate power, exacerbating inequalities.
2. Loss of Privacy: Personal data becomes a commodity in opaque AI-driven economies.
3. Erosion of Trust: Algorithms’ lack of transparency undermines societal cohesion.
4. Unintended Consequences: Competitive AI arms races escalate risks, from misinformation to autonomous weapons.

Harari’s warnings resonate with the challenges we face today. But what he misses is the potential of decentralised technologies to mitigate these risks, creating systems where AI agents can act transparently, equitably, and collaboratively.

By the way, I consider Nexus to be one of the most important books from the last few years and it should take a spot high up on the reading list of anyone who is interested in how technology is forming our future.?

A Vision for Multipolar Wins Through Decentralisation

Decentralisation is not just a constraint but an enabler of systems that scale equitably and ethically. Here’s how the building blocks of decentralised innovation can turn AI from a potential trap into a collaborative tool for progress:

1. Distributed Ledger Technology (DLT): Anchoring Trust

DLT provides the backbone for trustless coordination in AI ecosystems. By creating immutable records and enabling programmable smart contracts, DLTs have the potential to aid us in making sure AI agents act transparently and according to pre-defined rules. This addresses key issues like:

• Verification: Every AI decision or transaction can be logged and audited on-chain. Allowing for safety measures such as verifiable and automated oversight.
• Ownership: Tokenised assets and decentralised governance ensure that no single entity monopolises AI systems. The value of our data contributions can flow back to us, the community that owns, uses and enables the evolution of the AI itself.
• Privacy: Tools like ZKPs enable privacy-preserving verifications, ensuring that personal data remains secure while providing the visibility for community oversight and audit-ability.

2. Decentralised Autonomous Organisations (DAOs): Governing AI Collaboratively

DAOs offer a collaborative governance model:

• Democratising decision-making around the evolution of important AI that communities become dependent on, safe guarding the evolution of important AI from diverging away from the interests of the community.
• Programatic execution of fitness functions that are defined by the community and only allowing updates that perform at a level that represents an improvement in serving the community
• Agent-driven orchestration systems, where agents can represent human stakeholders, executing delegated tasks in full visibility and no exposure to incentives that don't serve the interests of the entire community.
• Align incentives through tokenised participation, ensuring that governance systems remain transparent, inclusive, and adaptive.

3. Zero Knowledge Proofs (ZKPs): Privacy Without Compromise

In a world where AI requires vast amounts of data to function effectively, ZKPs offer a solution to the privacy paradox. With ZKPs:

• Secure AI Training: Decentralised infrastructure and datasets can contribute to AI training without exposing data or opening security vectors. Allowing the value of AI training to be shared amongst the community unlocking the huge amount of redundant resource that we are all holding on our personal devices.
• Verifiable Interactions: AI agents can prove compliance or outcomes without revealing sensitive information or proprietary IP.
• Enhanced Trust: Users gain confidence that their data is being used ethically and securely.

4. Fully Homomorphic Encryption (FHE): Privacy-Preserving Collaboration

FHE is a groundbreaking technology that enables computation on encrypted data without ever exposing the underlying information. In the context of decentralised AI, FHE can:

• Enable Secure Collaboration: Multiple stakeholders can contribute their encrypted data to AI models without compromising privacy. For example a group of healthcare entities could contribute patient data to train a collective medical AI without breaching personal data privacy regulations.
• Federated Learning: AI models can be trained collaboratively across decentralised nodes without accessing sensitive information in the data.
• Privacy-Preserving Insights: Organisations can generate valuable insights from sensitive datasets while ensuring compliance with privacy regulations.

5. The Agentic Layer: Empowering Communities

At the heart of a decentralised AI vision is the Agentic Layer, where AI agents autonomously fulfill user intentions. These agents are goal-oriented, adaptive, and operate within trustless systems powered by DLT. Key benefits include:

• Scalability: AI agents handle complex workflows across decentralised networks.
• Resilience: By leveraging decentralised infrastructure, AI agents avoid single points of failure.
• Economic Participation: Users, through wallets and tokenised incentives, actively shape AI ecosystems rather than being passive subjects.

A Call to Action: Building Multipolar Wins

To move from Harari’s dystopian warnings to a more optimistic vision, we must act. Founders, technologists, and policymakers need to:

1. Open up the Conversation: Its important that this subject is not kept in a technological silo, everyone needs to understand and work together and the more we talk about this movement the more likely we are to achieve a future in which we harness the power of our technology for good.
2. Prioritise Decentralisation: Build AI systems that leverage DLT, ZKPs, FHE, and DAOs to ensure transparency, security, and inclusivity.
3. Focus on Usability: Develop UX that that bridges the gap between technical complexity and human engagement.
4. Embed Ethics in Design: Use tokenised incentives and agentic systems to align AI development with community values.

Conclusion: From Fear to Flourish

The future of AI doesn’t have to be a trap. By embracing decentralised technologies, we can create systems that amplify human potential while safeguarding against the risks Harari warns about. This isn’t just a vision—it’s a blueprint for a future where decentralised AI enables human agency enabling us to identify and create multipolar wins. Let’s not allow fear to paralyse us, let's get stuck in and build this future together.

Thanks

Me and my closest people have been talking about this problem space for almost a decade already. The term "multipolar wins" comes from a close friend and mentor Jerome Kelsey

A whole heap of the inspiration in working and thinking in this area comes from previous work with Pul Bandara

Enjoying some of the musing coming from Jamie Burke and the Outlier Ventures team in the Post Web series - looking forward to the subsequent chapters!

Obviously the main inspiration to this comes from Nexus so special thanks to the author Yaval Noah Harari.