I have a lot to say in response to the Future of Life Institute open letter on pausing AI development and will share more very soon, but TLDR: this approach will not work. You can’t stop an avalanche with a hammer, especially not a multibillion-dollar avalanche. We already have a tool to manage AI risk more proactively: blockchain. We also can’t over-determine or even accurately predict the effects of a technology before it happens. It’s an emergent phenomenon. For technology this complex to be constructive, we need to manage it from inside the system by way of governance and mechanism design. The future of AI is open source, governed by a decentralized community that can manage bias by guiding data labeling, set up safety and monitoring protocols, authenticate data, and share accountability in a more transparent way not controlled by a single company. The future of the internet is authenticated, and it includes powerful AI systems where every piece of information has provenance which is clearly displayed to the user. Information and requests made on users behalf are recorded and audited in tamper-proof ways. Today’s digital systems are huge, complex, interconnected, and evolving rapidly – centralization only increases the risks they present. Stopping them in order to guess what might happen misses the point entirely, and so does trying to fix something after it goes wrong. We already have the tools to set up proactive, open AI governance and data provenance today with blockchains, and specifically with the Blockchain Operating System: coordination tools, voting, and data authentication for all actions and content on the platform. The BOS is an operating system for a post-AI world.