Opinion of: Samuele Marro, machine-learning doctoral student at the University of Oxford
The AI web3 space fell into a trap. Somewhere along the way, “decentralized IA” has become synonymous with “IA blockchain”. This false equivalence actively harms innovation.
Excellent decentralized AI projects refer to blockchain executives, not because it has technical meaning, but because it is the only way to access funding, expertise and web 3 communities. The blockchain mentality does not only limit the decentralization of AI could become; He cannibalise it.
Web3 is not the blockchain
The ideals of web3 came out of Cypherpunk: without confidence, without authorization, resistance to censorship and property of users. The industry has forgotten a critical distinction: Web3 philosophy differs from Blockchain technology. Bittorrent is web3. Tor is web3. IPFS is web3. And now that web3 AI is at the center of the media, many find that blockchains are often poorly suited.
Jump in any AI web 3 space, and you will see the same model repeat: brilliant teams creating distributed learning networks, AI markets in Peer-to-Peer (P2P) and distributed training systems, explaining all awkwardly why they need a token or the functioning of their onchain colony.
As a counterexample, consider federated learning, where several nodes collaborate to form a shared AI model while keeping their private raw data. This is an excellent example of decentralized AI: no token is required.
This does not mean that the blockchain is never useful. Onchain’s regulations can simplify payments between AI agents, cryptographic evidence can improve reputation systems and tokens can align incentives in collaborative training. But these are specialized tools, not solutions to a single size. For many decentralized problems in AI, blockchain’s general costs only add latency, complexity and cost.
A game of incentives
Why do projects make these decisions? The answer lies in how the web 3 ecosystem has evolved. Projects that do not integrate blockchains are not considered “web3” and therefore cannot access web financing pipelines, expert networks or community resources. In their thesis, venture capital funds with “Web3” have established investment criteria around the integration of blockchain. Web3 AI spaces treat non -blockchain projects as out of reach.
These incentives push the teams to adopt the blockchain not for product reasons but for access to ecosystems. In other words, they make architectural decisions according to fundraising requirements rather than optimal user results. There is nothing wrong with playing the game, but that means that many opportunities for authentic (and profitable) applications of a decentralized AI are neglected.
The industry must recognize that three distinct concepts have been artificially grouped.
In relation: Can the blockchain provide the railings to keep AI on the course?
The decentralized AI includes distributed computers, federated learning, P2P networks and EDGE IT, which in accordance with Blocchain infrastructure intrinsically.
The AI integrated by Crypto implies token incentives, cryptographic evidence, management of digital assets and legitimate use cases which can be implemented using blockchains.
Web3 AI represents the possession of users, innovation without authorization and community governance, which can be carried out thanks to several technological approaches.
These concepts can work beautifully together, but they don’t need them. A federated learning network can use cryptographic confidentiality guarantees without touching a blockchain. A distributed AI market can implement validation based on reputation without intelligent contracts. Incituration systems can operate thanks to consensus mechanisms that do not require the general costs of an entire blockchain infrastructure.
Decentralized AI needs a toolbox
The real innovation in decentralized AI requires technological pluralism, blockchain being a tool in the toolbox, not a religious requirement. The most successful projects of the next decade will be those who choose the right architecture for their specific challenges, not those that comply with the current expectations of the ecosystem.
The financing of web3 and community support must evolve to adopt non blockchain decentralization. Capital funds can obtain substantial yields on decentralized and aligned projects on web3, even if their financing model is not based on tokens.
Communities must celebrate innovation without authorization regardless of its technical substrate. For example, many decentralized AI ecosystems exist beyond crypto, non-profit and profit. Prime Intellect has formed large models of large -scale language while preserving decentralization. The Nanda of the Massachusetts Institute of Technology builds an internet for decentralized agents. Laion democratizes research on AI.
These systems reach real decentralization; They do not wear blockchain badge and are invisible for a large part of the web3 community. Even in the more traditional web3 space, however, there are positive signals of projects that use block channels only when logical.
Nuirai uses the chain to manage the challenges for the models developed by the community, rewarding the most efficient. Torus Network transforms the tokens awards transparently to the agents who contribute most to its long -term growth while capturing the value of the network in the token. Payment based on Render Network tokens means that anyone, anywhere, can provide a calculation to its rendering farm. All these applications are already there.
The current approach of blockchain limits innovation in a decentralized AI precisely when it is most necessary. As AI systems become more powerful and centralized, there is a desperate need for decentralized alternatives. But they will not be carried out if the ecosystem continues to force each solution through the blockchain bottleneck. The projects that lose this ineffective state of mind today will dominate the ecosystem tomorrow.
The AI web3 faces a choice: to continue to cannibalize the decentralized AI with the requirements of the blockchain or release it to reach its full potential. Technology is ready. The question is whether the ecosystem is ready to evolve and which can capitalize on this change.
Opinion of: Samuele Marro, doctoral student in learning machine at the University of Oxford.
This article is for general information purposes and is not intended to be and must not be considered as legal or investment advice. The points of view, the thoughts and opinions expressed here are the only of the author and do not reflect or do not necessarily represent the opinions and opinions of Cointellegraph.
