It’s definitely artificial stupidity. I mean they’re never gonna win Jeopardy, but in important sense I think they were real creatures.
So I fell into this rabbithole.
Dr Steve Grand's life passion is to build a living AI, wether he's achieved it,(or wether he can) is a cause of confusion.
He seems to tentatively believe that they're alive, but tends to eschew that word, seeming to prefer the word "real." I think this has a certain meaning, that while they may not be alive, they are not "just machines" either.
His best known invention is the norn, a simple artificial life that were released to the public as pets in 1996 (as a part of a software called "Creatures"). I happen to be familiar myself. I believe they are smart enough to be stupid, as opposed to merely flawed.
Steve builds his AI on the basis of "bottom up design".
This means they should "start from the beginning" and be be built out of genes, neurones, and biochemicals (or simulated substances) that you then use to build a brain, organs, senses and learning (and theoretically thought, emotion). They should not know anything and should learn. Steve posits that living things are controlled by no instinct, impulse or ego, the entire thing makes the decision. That is "free will."
Humans tend to think that our ego (our "I,myself")controls us, this is opposed to Steve's definition of free will.
I believe for hardware AI (which Steve has) these rules function a little differently, I don't see how a steel thing would have genes or virtual organs, though Steve mentions hardware AI hold more promise because they can potentially see the world like we do rather than see a virtual one.
I could probably post a little more explanation on my understanding of Steve's work, but i'm already finding it hard to write this post in an easily readable way. Maybe ask me about it and I'll try to bring up one of his posts if I don't get it. Also I may make this a masterpost.
A note on the "emotional robots" you see on news articles. The goal appears to be to make them "display" humanlike emotions and play on the heartstrings of humans, as well as be directly programmed to peform functions. If the creator's intention is not a simulacra but something "real" then I want them to explain to me in painstaking detail why emotions cannot be merely programmed, because that's exactly what Steve does.
Steve Grand - Wikipedia
Wordpress - He....complains a lot, but his somewhat older posts about AI are here
Cyberlife-research - Archived website (old but good, explains stuff)
Creation - Book (read)
Growing up with Lucy - Book
The Brain in the machine - Magazine Article
Ever since C1 I’ve been trying to understand how the mammalian brain gives rise to imagination and mental imagery, and I think I’m really onto something now that can be implemented in real time on a PC (after all, computers are at least a thousand times more powerful than they were when I started writing C1). I think I have the key to an artificial life form that can actually think. Norns could react but they couldn’t think – they couldn’t make plans, have hopes or intentions, dream dreams, learn physical skills, etc. Higher consciousness can’t exist without an imagination either. It remains to be seen how smart they actually prove to be but they’ll certainly be much more realistic than the norns in lots of ways and hopefully a lot more fun to look after. - Grandroids