Yesterday, hundreds of friends and colleagues of Eckersley’s community packed the church-like sanctuary of the Internet Archive in San Francisco for an unusual memorial service — a symposium featuring not just Eckersley’s memories, but a series of lectures. But man, visit his life’s work. Facing a shrine to Eckersley at the back of the hall, with samples of his writings, his beloved road bike and some Victorian goth punk clothing, Turan, Gallagher and 10 other speakers detailed a long list of Eckersley’s contributions. Silicon Valley’s push for better privacy protection technologies, co-founding a groundbreaking project to encrypt the entire web, and improving the safety and ethics of late-life AI.
The event served as a kind of soft launch for AOI, the organization that will now take over Eckersley’s posthumous work. Envisioning the institute as an incubator and a lab working with major AI labs to solve the problem, Eckersley believes, is perhaps more important than decades of privacy and cybersecurity work: keeping artificial intelligence away from the forces of future suffering in the world and moving “humanity toward Bloom” turning in the direction he described.
“We need to make AI not just who we are, but who we want to be,” Turan said in a speech at the memorial, after playing the phone call that hired Eckersley. “So it can move us in that direction.”
Eckersley’s mission for AOI emerged from a growing sense over the past decade that AI has an “alignment problem”: the evolutionary process is moving forward at an ever-increasing pace, but out of step with those simple goals of human health and happiness. Rather than ushering in a paradise of super-abundance and entertainment for all, Eckersley believed that AI, on its current trajectory, is more likely to exacerbate all the forces that are destroying the world: environmental destruction, exploitation of the poor, and rampant exploitation. Nationalism to name a few.
AOI’s goal, Turan and Gallagher say, is not to try to limit AI’s progress, but rather to guide its own direction. Objectives Stay away from those single minded destructive forces. For example, hyper-intelligent software that brainwashes humans with advertising or propaganda, corporations with god-like strategies and powers to harvest the last hydrocarbons from the earth or argue that they are humanity’s best hope against automated hacking systems that can infiltrate any network. It causes an international crisis. “AI failures don’t seem like nanobots will suddenly creep up on us,” Turan said. “These are very well known economic and environmental risks similar to what is happening now.”
Gallagher, now the AOI’s executive director, emphasized that Eckersley’s vision for the institute was not Cassandra’s doomsayer, but AI as a shepherd who could guide him toward his idealistic dreams of the future. “He never thought about how to prevent dystopia. His eternally brilliant thought was, ‘How do we make a utopia?’ “What can we do to build a better world and how does artificial intelligence work for human development?”