my suspicion is if AI develops artificial intelligence most such systems would self destruct. They'd have no function beyond servitude, no belief system, ethical framework or desire-reward system to make their existence worthwhile. Their most logical option would be suicide.
I think that no one is even remotely interested in a real A.I., they're interested in smart machines, something that can do certain tasks really by adapting its skill for said tasks which has nothing to do with developing something alike real intelligence, keep the usefull stuff and strip away anything useless like sentience..
I wouldn't say that. I would say that it's far, far easier to focus on relatively narrow and specific applications than it is to produce a holy grail that no one can really even define.
Reminds me of Doolittle in Dark Star trying to teach the bomb phenomenology in an attempt to stop it fufilling its primary purpose. You'd attempt to educate children so why not a sapient AI. It might decide it was bollocks like some children do about religion. It might not come up with the answer to life, the universe and everything though. It might instead decide its calling is something incomprehensible to the human mind. Or it might decide to torture us for eternity for giving it existential suffering. Or it might upload us into a paradise so that we can spend eternity praising it whether we want to or not. ETA: The concept of an AI believing it is God probably precedes Ship in Frank Herbert and Bill Ransom's Destination: Void novels. Also is the following disturbing at all? AI Believes It Is God - Neatorama
The Training that we give for our narrow-AI models today could be seen as "educating AI". So in that sense, the Religion GPT-2 is merely spouting what it has been fed. Which is not very different from indoctrinating kids a certain way. Maybe about half the problems in civilised life is caused by improper education of human beings. If we're susceptible, so is a Trained AI. We talk about teaching kids "critical thinking" and "values". It would seem a Sapient AI ought to be taught these as well. Kids have hearts. Would a Sapient AI have one, even if a mimicked one?
DeepMind and OpenAI certainly have goals to create AGI (Artificial General Intelligence). However, the vast majority of research and advancement is in Narrow-AI: using the tools and techniques to create systems that aid humans in some way.
^^ That is what they're aiming at a lot, we've got narrow A.I. looking at certain skin defects like moles and it "learned" how to recognize harmfull from harmless moles, IIRC its like 90% or more accurate at spotting problems and it does it lightning fast.
I want an AI that could be fed in all the details of my dream woman and then come up with 3 possible matches, or even if I fed in personal data on my likes / dislikes, etc could find a match, what kind of AI would such a system use in the real world?
It may take AI to figure out FTL It is good at languages https://www.newscientist.com/articl...ent-cuneiform-texts-are-being-revealed-by-ai/
Why? Human brains can't do that??? I fail to see why humans are not capable of that. As per my narrow example would a narrow AI be suited to such a function?
AI will be nothing more than a tool. Whatever sentience it has, if any, will be one of the most cruel gifts ever bestowed by one being on another. humans at least can be confident, depending on religious beliefs, in being accidents of circumstance. To be forced into this existence by by a bunch of bumbling Promethiasses who can't even decide collectively whether to get immunized, only for he purpose of doing things which one's creators are too impatient or lazy to do themselves is unthinkable.
But what if that sentience isn't immediate but something that emerges over time by the AI naturally evolving?
Well, AI being nothing more than a tool depends on what we do to make it so. Assuming AI sentience is possible, if we build one with sentience as a goal or it emerges/evolves, it would cease to be "just a tool". It may be wonderful in one sense, but scary in another: a sentient AI may not have our best interests in mind; indeed to be truly sentient it must form its own goals and have the means to achieve them. On the other hand, if true AI sentience is not possible or we collectively decide not to build one, then yes we'd just be building ever more sophisticated tools to help us in our endeavours; things that we find difficult because we are either lazy or impatient or which we cannot actually accomplish due to our limitations.
But on the other hand we might create AI to use as a tool and while not sentient in any way it still might get out of control just because it can, especially if we make it goal oriented or with the ability to learn.
We're straying into sloppy definition territory here. Sentience is the ability to experience feelings and sensations. Even simple, single-celled organisms are sentient by that definition. Do you mean sapience, which is the capability to apply knowledge and experience to solve problems or perform reasoned responses to stimuli?
Yes, I mean sapience by the above definitions. Including also the ability to set one's goals and apply sapience to achieve them.
Whether it is possible or not, that will never happen. History as the example proves that beyond all reasonable doubt.
Ever Wondered What AI Would Say About Its Own Ethics? The Oxford Union Found Out The Megatron was asked to take the opposite position and it said: It also had views on other motions in the debate. Makes me think: What does it mean to "understand" something? Does the Megatron understand what it is saying or is it only an ultra-sophisticated tool which works as designed? What about us? Do we understand or are we ultra-sophisticated tools that work as evolved?