“Artificial general intelligence” (AGI) is typically defined as any computer system that can match or surpass human intelligence in performing any task a human can perform. No such program yet exists, but if one were to arise it could instantly begin improving its own capabilities. The result might be a kind of superintelligence, as far beyond our own intelligence as ours is beyond that of snails There exists no natural limit to this process. The only guardrails would be those we construct now, before the avalanche begins.
The timeline remains uncertain, yet the “San Francisco consensus” among AI researchers predicts superintelligence by decade’s end. Skeptics raise legitimate concerns about decades of failed predictions. But when Nobel laureates warn of extinction and industry leaders purchase remote bunkers or speak of “summoning the demon,” prudence demands attention. These are not Luddites but AI’s very architects sounding the alarm.