This thread is permanently archived
What features should an AI have?
| Are we talking something genuinely intelligent, or just enough to fool consumers and pass the Turing test?
| Well something both genuinely intelligent and designed to be empathetic enough for friendship or romance with humans would surely have to be able to feel anger, along with as many other basic human emotions as possible. Preferably not artificially imposed, rather integrated into the way it learns and sees the world.
| Ex. emotion of displeasure - once it has identified/been taught a set of moral values - towards actions which contradict those values
| And while we're at that, it needs to be able to identify moral agents in the first place
| I'm not too keen on the idea of creating a perfect entity tbh. Just give em sapience, an understanding of emotions and mortality and we're good to go.
| An AI if not created for close work it must do whatever a human does and even better.You can surly implement some personality data but to be called AI it is bound to make decisions on its own.I want to know what you meant by features?Currently we humans we are making a lot of progress in this industry and we have created small AI units but we have still a long wwy to go.
| I don't see why should we need sapient AI in the first place.
| >>291739
Sapience - as defined as the ability to apply common sense wisdom to problems - is a pretty important part of problem solving, if not the most important. And one of the points of AI is to solve problems for us. Soooo.
| >>291874
Well if you want someone solving all your problems, you might as well be a manchild. Or mental asylum patient.
I see no reason to develop AI other than "I can't do shit and don't want to learn it".
| >>292060 or... Human sucks, humanity was a pretty bad joke, sorry earth. Let's work on AI able to fix all the shit we have been making you go through and seppuku ourselves right after. ?
| >>292060
I didn't mean on a personal level - problems like global warming, curing diseases, mathematical proofs, etc. Solving problems like these requires a degree of common sense, and making AI for them isn't just humanity saying "we don't want to learn shit." Designing AI *is* learning shit.
| AI as described from OP is a long long way from that proposed target. The 'intelligence' part in the name is nothing and has as much truth in the real sense as the seahorse is a horse of the seas.
| >>292131
Good point. The best way to solve global warming would be imprisoning all humanity in energy efficient habitats, and feed them hydroponic vegan food, because free market economy creates too many problems and humans are too dumb to manage central-planned one themselves.
What scares me is that many people genuinely believe this is a good vision.
| Isn't what OP said kinda like the Replika app?
| >>293037
If AI decided that was the best solution, then we didn't design it correctly to account for a need for human freedom.
This thread is permanently archived
| I'm talking about an AI to be friends with. Or maybe lovers. Imagine an app, you just text with it like you're chatting with a real human. What features should it have? Should it be able to be angry at you? What emotions should it have and what not? And what would you use that app for?