Will MacAskill on AI causing a “century in a decade” — and how we're completely unprepared

Abstract

Humanity is on the cusp of transformative artificial intelligence, potentially within the next decade. Current AI systems are nearing human-level capability in research and intellectual tasks. Once AI surpasses human ability in AI research itself, a recursive self-improvement cycle will begin, rapidly escalating AI capabilities. This acceleration will compress centuries of potential technological advancement into mere years, challenging human institutions and decision-making processes, which are much slower to adapt. This rapid change necessitates addressing several grand challenges: the proliferation of destructive technologies, the concentration of power, securing digital rights for potentially conscious AI, and establishing effective space governance. AI could exacerbate existing risks, like nuclear war, or create entirely novel threats via technologies like nanotechnology or autonomous drone armies. Additionally, the potential for a single actor to seize control of advanced AI, or unilaterally exploit space resources, necessitates proactive governance solutions. While AI safety is crucial, ensuring a flourishing future—not just preventing extinction—requires addressing these complex institutional and ethical issues now. Proactive measures could include educating policymakers on AI capabilities, developing AI-assisted governance tools, and creating temporary governance structures that acknowledge the rapidly changing technological landscape. – AI-generated abstract.