Nillion could enable machine learning (ML) training and inference services to be run on decentralized private data at speeds significantly faster than incumbent tech competitors. By enabling the privacy-preserving computation of diverse high value data that was previously siloed in corporations, transformational changes in the fields of genomics, IoT, financial services could be unlocked.
NMC could also provide the backbone of secure and privacy-preserving LLM inference. Users’ weights of a model can be stored in NMC, facilitating other third-parties to use their weights for their own models without being able to reconstruct the training dataset nor identify the generator of the model weights. The ITS secure storage and sensitive, personal weights can protect user privacy with quantum-resistant security.
Source : https://www.nillion.com/ (tab : Artificial Intelligence and Machine learning ; tab located at the bottom of the nillion website home page)
I think there is a very large and very, very important overlap in terms of societal function between NMC and AI (inference). At the end of the day most people will have personal AI assistants to assist them with everyday life, which will include using their personal and private data. It would be ideal, optimal even, to ensure that data is not actually visible to the AI model, given that there could be issues or data leaks, or even LLM prompting attacks on it.
correct if we reach this stage where any individual have his personal AI being able to be sure that data are safe and not share with anybody will be so important. Because this personal AI will know everything about us as it will manage so much stuff in our life. So in this case having the service provider of the AI able to see you data is not acceptable !