From: Billy Brown <email@example.com>
<extracted from a conversation about the arrival of SI's>
>I say 'it', because I expect human-equivalent hardware to arrive several
>years before human-equivalent AI. That means the first self-enhancing AI
>that doesn't bottleneck will have the hardware to go from transhuman to SI
>in a matter of days, at most. At that point humans are far to slow to
>interfere with it - it can invent general-purpose assemblers, use someone's
>automated lab equipment to build one, and migrate itself to rapid
>infrastructure in a matter of hours. A few hours after that we've got a
>full-grown Power on our hands.
Why do people assume that an AI will arrive fully formed and ready to take on the world. Looking at the human mind, it takes us years, if not decades, to understand the world around us and become adept at manipulating it in complex ways.
The first AI may well be highly intelligent and a quick learner, but it will also almost certianly have to be spoon fed knowledge in its earliest stages and should be spottable as it 'grows up' . The chances of an AI spontanteouslycoming into being fully formed seems pretty low, as does the chances of an AI coming into being at a low level and increasing it's intelligence without tipping off the people around it (imagine managing to grow up from a baby to teenager whilst fooling the people around you into thinking that you only had baby-level intelligence - Especially if you didn't know that you had to make them think this.