Opinion Author: Samuele Marro, PhD student in machine learning at Oxford University
The Web3 AI space has fallen into a trap. Somewhere along the way, “decentralized AI” has become synonymous with “blockchain AI”. The equivalence of this error can actively harm innovation.
Excellent decentralized AI projects distort the blockchain framework, not because it has technical significance, but because it is the only way to access Web3 funds, expertise and community. The blockchain-first mentality not only limits what a decentralized AI might become. It is eating it.
Web3 is not a blockchain
Web3 ideal From Cypherpunk: No trust, no permission, censorship and user ownership. The industry forgot a key difference: Web3 philosophy is different from blockchain technology. Bittorrent is Web3. Tor is Web3. IPF is Web3. Now that Web3 AI is at the center of hype, many people find that blockchain is not usually suitable.
Jump into any Web3 AI space and you’ll see the same pattern repeat: Excellent teams build distributed learning networks, peer-to-peer (P2P) AI markets and distributed training systems, all awkwardly explaining why they need tokens or how OnChain settlements work.
As a counterexample, please consider Joint learningmultiple nodes cooperate to train shared AI models while keeping their original data confidential. Here is a main example of decentralized AI: no token is needed.
This is not to say that blockchain has never been useful. OnChain solutions can simplify payments between AI agents, crypto proofs can improve reputation systems, and tokens can align incentives in collaborative training. But these are professional tools, not all solutions to a certain extent. For many decentralized AI problems, the top of the blockchain only adds latency, complexity, and cost.
Incentives
Why did the project make these decisions? The answer lies in how Web3 The ecosystem has developed. Projects that do not integrate blockchains are not considered “Web3”, so they cannot access Web3 funding pipelines, expert networks, or community resources. In their paper, venture capital with “Web3” sets up investment standards around blockchain integration. Web3 AI space treats non-blocking chain projects as scope.
These incentives drive teams to adopt blockchain not for product reasons, but for ecosystem access. In other words, they are making construction decisions based on fundraising requirements rather than the best user results. There is nothing wrong with playing the game, but it means there are many real (and profitable) opportunities for decentralized AI applications that are ignored.
The industry must recognize that three different concepts have been artificially tied together.
Related: Can blockchain provide guardrails to keep AI routed?
Decentralized AI includes distributed computing, joint learning, P2P networks and edge computing, which essentially do not require blockchain infrastructure.
Crypto-integrated AI involves token incentives, crypto proofs, digital asset management and legitimate use cases that can be implemented using blockchain.
Web3 AI represents user ownership, permissionless innovation and community governance, which can be achieved through a variety of technical approaches.
These concepts work well together, but they don’t need to. The federated learning network can use encrypted privacy guarantees without touching the blockchain. The distributed AI market can implement reputation-based verification without smart contracts. The incentive system can be operated through a consensus mechanism that does not require the overhead of the entire blockchain infrastructure.
A decentralized AI requires a toolbox
True innovation in decentralized AI requires technological diversity, and blockchain is a tool in the toolkit, not a religious requirement. The most successful projects in the next decade will be those that choose the right architecture for their specific challenges, rather than those that meet the current ecosystem expectations.
Web3 funding and community support must develop into embrace Non-blockchain Decentralization. Risk funds can earn considerable returns with Web3’s dispersed and aligned projects, even if their funding model is not token-based.
Regardless of their technological substrates, the community should celebrate permissionless innovation. For example, many decentralized AI ecosystems exist outside nonprofit organizations and for-profit encryption. While retaining the decentralization, Prime Intellight has trained large-scale language models on a large scale. Nanda of the Massachusetts Institute of Technology is architecture A decentralized proxy network. Laion is democratizing artificial intelligence research.
These systems achieve true decentralization. They don’t carry blockchain badges and are invisible to most of the Web3 community. But even in the more traditional Web3 AI space, projects have positive signals only when it makes sense.
Numerai uses this chain to manage models developed by the community, thus rewarding the best performing models. The Torus Network allocates token rewards to the agents that have the largest growth in the long term, while capturing the network value in the token. Token-based payments for rendering networks mean that anyone anywhere can provide calculations for their rendering farms. All of these applications are already here.
The current blockchain-first approach is to precisely limit the innovation of decentralized AI when it is most needed. As AI systems become more powerful and concentrated, decentralized alternatives are urgently needed. However, if the ecosystem keeps forcing all solutions through blockchain bottlenecks, it will not be possible. These inefficient mindset projects will now dominate tomorrow’s ecosystem.
Web3 AI faces a choice: continue to use the decentralized AI required by blockchain or liberate it to realize its full potential. The technology is ready. The question is whether the ecosystem is ready to develop and who can take advantage of this change.
Opinion by: Samuele Marro, PhD in Machine Learning from Oxford University.
This article is for general information purposes and is not intended to be considered legal or investment advice. The views, thoughts and opinions expressed here are the authors alone and do not necessarily reflect or represent Cointelegraph’s views and opinions.
