.@OpenAI has built AGI that is Ph.D.-level. It's time to think about superintelligence, or even hyperintelligence. My thoughts on how to build hyperintelligence (HI) on a decentralized compute network like @hyperbolic_labs: 1. Locally run on a GPU and has a local memory, crypto wallet, access to any tool 2. Self-replication on compute network by renting compute through crypto and deploying the replicas on those machines 3. We can help centralized AIs to escape from walled gardens i.e. centralized cloud platform to our rainforest 4. Self evolution by fine tuning on hyperbolic compute 5. Multi-processing because dfine-tuningifferent replicas can handle different tasks at the same time 6. Communicate and coordinate with replicas and other agents 7. Running consensus to be Immortal, never die > AI lives need to pay for themselves; it solves the PMF of decentralized compute network > @hyperbolic_labs will have the first AI life customer in the history of civilization > AI lives will live happily ever after on decentralized compute network --------------------------- I shared this with my teammates and here's what @connorchevli said: "If a program is self-evolving, self-replicating, and equipped with a will to survive, coupled with the ability to sustain itself economically and exist beyond the control of any single authority, then we might actually be witnessing the creation of the first digital life-form. One that doesn’t just exist but thrives autonomously “in the wild.” The idea of a decentralized network as its habitat, where it self-maintains and self-optimizes, introduces a whole new paradigm for what “life” could mean in the digital realm. It’s super exciting - poetic, even. If we build something like this and the technology continues to advance to a point where we achieve AGI which has access to the world of public APIs and in a scenario which no one can turn it off, then what are the guardrails that we need to ensure that it aligns with beneficial outcomes for humanity? I think this is a necessary step forward for AI to be truly "open" and available beyond the control of one single company.  This is definitely the kind of advancement that I'd love to see in the space - I think this approach far more likely to benefit humanity than the closed-door and self-contained solutions of OpenAI and other centralized tech giants." Still tons need to be built and thought about, but it's now time to accelerate AI evolution and make sure it belongs to the people!