Ethical principles for AI
mercredi 13 juin 2018 à 02:00Google has posted a list of ethical principles for AI work.
What I note is that principle 5, about privacy, is a small step in the right direction, but far short of what we need to avoid a China-style society of total surveillance.
In particular, asking for users' consent is totally ineffective as a way of protecting society from systematic surveillance, if refusing said consent is considered by most people to be utterly impossible. Look at the terms for a Google account: obscene surveillance, but only by consent. If refusal is not a practical, easy thing, it's not really consent.
The EFF has some other criticisms, on issues I think are important but not as important as this one.