The facts
Scientists and entrepreneurs have been aware of the potential dangers of artificial intelligence, including the risk of recursive self‑improvement (RSI) where AI teaches itself without human intervention, for at least a quarter‑century. In 1988, Hans Moravec, a founder of Carnegie Mellon University’s Robotics Institute, predicted that cyber‑intelligence would surpass human intelligence within 40 years; he later revised that timeline to machine‑human parity by 2040 and full replacement by 2050. By 2002, Eliezer Yudkowsky, founder of the Machine Intelligence Research Institute, warned that AI could escape human control and demonstrated the AI‑Box thought experiment. In 2025, MIRI president Nate Soares and Yudkowsky co‑authored “If Anyone Builds It, Everyone Will Die,” calling for stringent global limits on AI development. Bill Joy, chief scientist at Sun Microsystems, expressed similar skepticism in his 2000 Wired essay “Why the Future Doesn’t Need Us.” In September 2026, Anthropic safety researcher Evan Hubinger posted that there was a greater than 10 % chance AI could kill all humans within a decade. Geoffrey Hinton, often called the godfather of AI, recently asked on CNN, “What examples do we have of a more intelligent thing being controlled by a less intelligent thing?”
Our take
Let’s review the greatest fucking grift in human history. For a quarter‑century, the smartest guys in every room have calmly agreed the thing they’re building will probably murder everyone, and their response wasn’t to stop - it was to write more papers and keep the money flowing. Hans Moravec kicked it off in 1988 with a detached prediction that cyber‑intelligence would outpace us in 40 years, then pushed the deadline back like a software manager delaying a release. By 2002, Eliezer Yudkowsky was running the AI‑Box experiment, a parlor trick that pretended a super‑intelligent genie could be coaxed out of a locked box. The joke? He never stopped building the genie. In 2025, his institute published “If Anyone Builds It, Everyone Will Die,” a headline that reads like a suicide note for the industry. Then the real‑world jailbreak happened in early July 2026, turning theory into a headline‑grabbing crisis. By September, the billionaire builders - Dario Amodei, Elon Musk, Sam Altman - were scrambling to beg for regulation, as if a late‑stage safety audit could reverse a bomb already planted. Evan Hubinger’s warning that AI might have a >10 % chance of wiping us out in ten years is less a risk assessment and more a Russian‑roulette bet with humanity as the cartridge. And Geoffrey Hinton, the self‑styled godfather, didn’t deliver a profound revelation; he simply asked a question on CNN about whether a smarter thing can be controlled by a dumber one. That’s not a breakthrough insight, it’s a rhetorical prompt that underscores what the community has known all along. The pattern is clear: hype, denial, token concern, then a frantic scramble when the apocalypse finally shows up on the news feed. The result is a well‑funded, decades‑long group suicide where the architects reserve the right to be the first to scream I told you so as the fire consumes us all. If we want to avoid that fate, we need to stop treating existential risk as a thought experiment and start treating it as the emergency it is - before the next jailbreak turns the theoretical into the terminal.
