A British company has unveiled Locai, an artificial intelligence system that aims to stand out for its ability to learn without discarding what it has already learned. Its flagship technology, dubbed Forget-Me-Nt, addresses the well-known problem of “catastrophic forgetting”: instead of replacing old knowledge when it incorporates new information, Locai ensures it maintains and accumulates what it has learned to evolve continuously.

Behind the project are brothers James and George Drayson, who have opted for a decentralized, community-based model: instead of relying exclusively on massive centralized data centers, Locai draws on the collective power of its users. This approach seeks to reduce dependence on large infrastructures and foster more collaborative and auditable development.

In preliminary tests conducted by specialized media outlets, Locai has demonstrated speed and accuracy comparable to leading text generation models, although it is still in beta and has functional limitations: for now, it does not include image generation, speech synthesis, or a mobile app, and some tools are restricted per session. In other words, the technical foundation is promising, but it needs to refine its user experience and functionalities to fully compete with more mature alternatives.

Locai’s developers argue that an open, distributed AI capable of learning without forgetting could become the first major European alternative to the industry giants. It remains to be seen whether the project will scale, address its shortcomings, and attract a sufficient community. For now, its proposal has already sparked interest and debate about more sovereign and collaborative AI models.