“There is no theoretical barrier to constructing human level AI’s, since humans’s exist. We can improve human intelligence in various ways (slightly), so we can improve these AI’s too (so slightly better than human is possible too). For super-AIs, on the other hand, nothing exists in the world to show that they are possible.”
If the “man on the street” has sufficient familiarity with The Wonders of Science to accept human-level AIs as genuinely possible, it seems to me that the sheer boost in simple processing power available to an artificial construct is probably enough to support at least a theoretical acceptance of the possibility of super-AI. If somebody understands technology well enough to allow for near-human AI, I would expect to find that they assume super-AI is not just possible, but trivial … witness people attributing all sorts of mysterious and nefarious intelligent behavior to Google’s on-the-fly search predictions, for example...
“There is no theoretical barrier to constructing human level AI’s, since humans’s exist. We can improve human intelligence in various ways (slightly), so we can improve these AI’s too (so slightly better than human is possible too). For super-AIs, on the other hand, nothing exists in the world to show that they are possible.”
If the “man on the street” has sufficient familiarity with The Wonders of Science to accept human-level AIs as genuinely possible, it seems to me that the sheer boost in simple processing power available to an artificial construct is probably enough to support at least a theoretical acceptance of the possibility of super-AI. If somebody understands technology well enough to allow for near-human AI, I would expect to find that they assume super-AI is not just possible, but trivial … witness people attributing all sorts of mysterious and nefarious intelligent behavior to Google’s on-the-fly search predictions, for example...