Impromptu: Amplifying Our Humanity Through AI For example, when my co-founders and I launched LinkedIn twenty years ago, we were motivated largely by the fact that the divisions between “cyberspace” and “the real world” were rapidly collapsing. Instead of existing as a place that people “went to” under the cover of pseudonymous screen names, the internet had evolved into a place that people were using to facilitate their lives. They went there to buy things, and stay in touch with their families, and make plans with their real-world friends. In such a milieu, my co-founders and I realized, digital platforms built on real identity could be hugely beneficial to hundreds of millions of people. In LinkedIn’s case, of course, we focused on professional iden- tity. To create trust on our platform, we positioned a user’s identity in networks of affiliation. That way, you couldn’t just craft a fictional persona out of thin air—other users effectively confirmed that you were who you said you were. At Facebook, launched a year later, participation was initially limited to students with verified college email addresses. And while Facebook didn’t require new users to upload photos of themselves, that’s clearly the norm it was hoping to establish. Through such approaches, social media helped real identity take root online in a meaningful way for the first time. And yet social media’s emphasis on broad participation also left it vulnerable to various kinds of deception, some human-gener- ated, some automated. As I’ve written elsewhere, I believe most social media platforms underestimated how much governance online communities would require, especially as they scaled to hundreds of millions of users or more. 102