We believe that the only way to have people trust AI is when we will make it possible to have privacy-respecting AI, we believe it is not only feasible, but it is our duty to build AI which respects people and is able to scale without presenting a danger to our societies
Same. Got excited, then saw it's shilling a coin [1]. From a branding perspective, this undoes the goodwill won by the privacy message.
I know about Snips.ai for some time now, I was planning to play around with it soon, and then I learned about the token thing. I mentally put the project into the trashcan at that very moment, it only got out of there due to being repeatedly mentioned on HN as something legit.
They're promoting a security [1]. If a company can't bother with the law, I distrust their claims around privacy.
Also, if you're doing on-device AI, you have a natural business model: sell the device. (Maybe also subscriptions.) Adding in a token creates conflicts of interests.
[1] https://www.sec.gov/oiea/investor-alerts-and-bulletins/ib_co...
And when we consider all this, we are thinking about:
- how can we remunerate people who want to share their data
- how can we ensure that people can buy apps without leaking information about who they are and what they do
- how can we incentivize the creation of an encryption-based machine-learning network which would allow to do private-by-design model training
and the answer to all of these was too build a decentralized AI
you can learn more about this on our blog:
- https://medium.com/snips-ai/decentralized-machine-learning-u...
- https://medium.com/snips-ai/token-curated-app-stores-dec6a0c...