Collective and distributed ownership models for AI


Hi all - Does anyone have any knowledge of collective and distributed ownership models for AI? A contact mentioned that Nesta may be opening some funding in the future on a similar topic.

Would be good to learn more about it - maybe there’s a potential collaboration too if such funding opens up?



The Mycroft AI project - - has an investment offer running curently where they are suggesting that people can become a “community partner”. In reality it is non-voting shares that they are selling. Ownership without a voice.


There’s been quite a lot of conversations about whether this would help tackle issues of inequality (of wealth and power) in AI development. One question is whether the ownership is of the datasets (used to train and develop the AI) or the system itself, or both. Public good / commons models are increasingly coming to the fore in discussions about data and AI (I mentioned this in evidence to the APPG on AI last year). Last year’s report on AI proposed data trusts - one of the interpretations of this is some sort of mutual (recent attempt to detangle confusion here from the ODI - ). Edge computing techniques and greater control of personal data suggest a future with distributed machine learning operating in a privacy-preserving fashion; not sure anyone is close enough to deployment to think about ownership models? I’m not aware of any current instances of shared ownership, but would be glad to hear of them :slight_smile: (there was debate about whether gig economy workers, used to train AIs in some cases, should have rights over the AI value created; I often think about that when doing Google’s image recognition work through captchas…)