Virtual assistants, such as Amazon’s Alexa, the Google Home, and Apple’s Siri are now being used by millions of users, every day. As they have become more sophisticated, people have begun using them for many kinds of things beyond checking the weather: shopping, scheduling, time keeping, among others.
As these devices become more central to people’s lives, they start to get privileged access to many private aspects of people’s lives: not only access to their private spaces, such as what is said at home or in the bedroom, but also the kinds of things people do and buy. Colluding with smartphone apps, such devices will soon be able to infer your emotional and physical state, your plans and intents, your income, your conditions, and your personal activities.
How do we know these virtual assistants are genuinely serving their users with their best interests in mind? How do we also know what the companies behind these devices will do with your data?
As we have written previously about “respectful AI”, this is not just about privacy, although privacy is an important element. This is about whether such assistants genuinely serve you with your best interests in mind. Will a large online retailer, such as Amazon, always design its virtual assistant to get you only the things you need via the best means of getting them–or might they, for example, design their virtual assistant to make them more money, such as making you want more things?
We think that, in the future, the growing importance of virtual assistants will underscore the need for neutral virtual assistants for and by the people. By “by the people”, we mean non-retailers, non-advertising companies, but free an open source developers who genuinely are neutral and have users’ best interests in mind.
Towards that end, we have taken the first step with ARETHA (* Artificial Respect-Enabled Transparent Home Assistant) project a FOSS Virtual Assistant platform for experimenting with respectful behaviours using DIY hardware for a smarter home.
The prototype is under development and not ready for public use. Check out the GitHub if you’d like to contribute!