A team of scientists in Belgium may have solved one of the biggest challenges in AI using a blockchain-based, decentralized training method. While the research is still in its early stages, the potential implications range from space exploration to existential threats to humanity.

In a simulated environment, the researchers developed a way to coordinate learning among independent, autonomous AI agents. Blockchain technology was used to facilitate and secure the agents’ communications, creating a decentralized “swarm” learning model.

The individual training results from each agent were used to train a larger AI model. Because the data was processed via the blockchain, this larger system benefited from the collective intelligence of the swarm but did not access the data of individual agents.

The research team conducted their blockchain research using a learning paradigm called “decentralized federative learning.” This allowed them to coordinate their models while maintaining data decentralization.

The team studied the swarm’s resilience to various attack methods. Since blockchain technology is a shared ledger and the training network used in the experiment is decentralized, it has been shown to be resistant to traditional hacking attacks.

However, it was found that there is a strict limit to the number of malicious robots that the swarm can handle. The researchers developed scenarios that include agents designed to disrupt the network. Simple and outdated agents are relatively easy to defend, while intelligent agents equipped with malicious agendas can disrupt the swarm intelligence.

This research is experimental and has only been conducted with simulations. However, there may come a time in the future when AI agents from different companies or countries can work together to train a larger agent without compromising data privacy.

Share your opinions in the comments!