Bittensor review: has the synergy of blockchain and artificial intelligence arrived?

With the launch of ChatGPT, the first massively artificial intelligence (AI)-based product, the lives of millions of users have been divided into "before" and "after". Even this relatively limited solution has transformed numerous fields and businesses, including the cryptocurrency industry. 

During the hype surrounding ChatGPT in 2023, a segment of projects attempting to combine blockchain with AI emerged in the crypto market. And although most of these startups are just riding the hype wave, the synergy of decentralized networks and artificial intelligence can actually solve a number of important technological problems.

The MoE concept

"The Holy Grail" of the AI industry is to create a so-called general artificial intelligence (AGI) that, like a human, could solve diverse tasks and learn flexibly. However, this is an extremely difficult task even for technology giants with their virtually limitless resources.

Bittensor as an alternative uses MoE (Mix of Experts) architecture, which involves creating a network of specialized AI models. This network processes the user's request and produces the highest quality answer from an "expert", i.e. the most relevant algorithm. 

Comparison of centralized and decentralized AI model. Data: Revelo Intel.

In practice, this is implemented using subnets specialized for specific groups of queries. Any user can register his subnetwork for a certain fee and attract AI algorithm providers to it to solve a particular task, receiving 18% of the total income of the participants. But it is important to note that the maximum number of subnets is set by developers, so in this direction Bittensor is still not a fully open blockchain.

So, at the time of writing, Bittensor's network consists of 32 subnets, and among the main specializations are generative art, translation, text queries and others.