Arcam AB fka AMAVF

[Tom Joad]

re: intelligent computer safeguards

Now there's a can of worms! There's no anser. I would say now that everything below is off topic, so I'll send out only one reply.

A fairly full description of why there's concern-

AI revolution

So far, thankfully, computers are pretty stupid and depend on our programming skills. They do poorly at common sense knowledge of the world around them, only recently has a computer been trained to see the difference between a dog and a cat, and they're terrible when it comes to doing anything that's not rule based, say, an android cracking an egg to cook in a skillet.

So Deep Blue beat Kasparov at chess? Here's how that went, Deep Blue had a library of all past chess games in its memory banks. Deep Blue could calculate what followed by moving every legally movable piece in a timely manner. Kasparov was looking at what followed from moving one or two pieces. The results weren't so one sided as most believe. There were 3 draws out of 6 games in both matches in 1996 and 1997.

Computers have a set architecture with adaptive software, but we have a fluid architecture since we continue making new neuron connections while alive, and we also with "fluid software," as we can learn different problem solving strategies.

Computers have a difficult time taking optical data and deciding what defines an object. Example: Tesla crash in Florida when the self driving car couldn't distinguish the side of a truck from the sky (or so says Musk). I know the driving data for self-driving cars is good, but think of the problem of taking in color, color difference, and distance data and defining an object recognition algorithm. We can be tricked by a magician but we're pretty good at taking optical data and then knowing where the objects are.

A more in depth description of why some aren't concerned-

Hubert Dreyfus views on AI

If you haven't seen it, time to see Bladerunner, because if we get there we have to ask if it's ethical to create a conscious being.