Can AI recommendation algorithms provoke new interests instead of recycling familiar preferences?
AI recommending is now stretching across human experience from entertainment (Netflix, Spotify) to society and media (X, Facebook) to professional life (LinkedIn, Indeed), to healthcare (Woebot, Headspace), to personal life (Tinder, Bumble).
Increasingly, AI Recommendations are not only facilitating choices, but also allowing choice to exist in the first place. When everything and everyone and every experience is available online, the superabundance is overwhelming. There is no way to make sense of the possibilities without an initial filtering reducing the options to manageable dimensions.
Consequently, AI recommendation algorithms condition the possibilities of what we do, where we do it, with whom, and why. As individuals and as a society, the algorithms are no longer helping us to choose so much as shaping what can be chosen. We no longer use them so much as they define us.
Recommendation algorithms are the keystone problem of the AI human condition, they determine what it means to be human. So, the dilemmas of tuning and weighting recommenders are more significant than earlier preoccupations with privacy, autonomy, fairness, and explainability. The ethics of recommenders is more significant than all the rest put together because their questions are about who it is that will protect privacy, and who will be free, and who will uphold fairness, and who will search for explanations.