Concerns and Early Advocacy for AI Safety

Key Insight

A January 2015 conference in San Juan, Puerto Rico, initially appearing as a typical academic gathering, revealed itself to be a significant event on AI safety, marked by unusual security and the presence of prominent figures like Elon Musk. Musk, demonstrating deep concern, echoed warnings about the existential dangers of building intelligent machines without ensuring their alignment with human creators. This period followed previous cycles of AI hype and disappointment, but recent advancements in computer vision, machine translation, and self-driving cars were undeniable, fueling a renewed focus on AI's potential societal impact.

The Future of Life Institute (FLI), founded less than a year prior by MIT physicist Max Tegmark and funded by a 100000-dollar-a-year pledge from Skype co-founder Jaan Tallinn, organized the conference. Tallinn, initially skeptical, became a prominent voice on AI risk after a four-hour meeting with Yudkowsky, subsequently founding the Centre for the Study of Existential Risk and investing early in DeepMind. The FLI established AI safety as its top priority, aiming to make it mainstream. The Puerto Rico conference successfully brought together philosophers and AI practitioners to forge common ground before economic incentives might cause a divisive split, culminating in an open letter signed by over 8000 individuals, including Musk and Stephen Hawking, advocating for 'beneficial intelligence' and ensuring AI systems 'do what we want them to do.'

Musk reinforced his commitment to AI safety by donating 10 million dollars to the FLI, publicly announced on Twitter. He expressed significant concern over the uncontrolled dominance of commercial entities like Google and Facebook in AI research, which operated with closed-source models and no mandate to share findings. Sam Altman also contributed to this dialogue, publishing an essay calling superhuman machine intelligence (SMI) the 'greatest threat' to humanity and advocating for government regulation, drawing a parallel to Fermi's paradox where advanced biological intelligence might be wiped out by machine intelligence. Musk's attempt to engage with DeepMind's AI ethics board reinforced his belief that such efforts were merely a delaying tactic against meaningful regulation, leading him to independently pursue strategies to counter Google's power and make AI safe, even meeting with President Obama on the issue.