http://plato.stanford.edu/entries/zombies/
Logically, one cannot determine if they are a zombie, but of course qualia does not deal with logic.
Should AI be regarded as Turing-test verified?
- Thread starter Kyriakos
- Start date
http://plato.stanford.edu/entries/zombies/
Logically, one cannot determine if they are a zombie, but of course qualia does not deal with logic.
From what I've seen so far, I'm not especially impressed with AI ability to simulate humans. That said, I'm also not sure that the point at which they do manage to simulate humans convincingly in text conversation will be very significant from a practical angle.
However, it will make online spam much more irritating. Right now, if a spambot shows up on CFC, it's pretty obvious. Most commonly, there will be poorly-written drivel of some sort leading to one or more links, usually advertising free movies, some website, cheap v1@gr@/c1@l1s, etc. Alternatively there could be an innocuous-seeming comment like "i totally agree with that" followed by a tracking pixel, or such a comment could be posted without a tracking pixel or link, and then if left alone the bot will either come back and edit in a link later on, or it may make a few of those posts and start in on the links only after a few innocuous posts. It can also copy random text from an earlier post and paste it, which at least looks more human because a human wrote the original.
At present, that's about all we see, and it all looks pretty stupid. A Turing test-passing spambot, however, could be far better. They could be written to act like normal posters for their first hundred or so posts, giving sensible replies to a variety of threads and then slowly beginning to introduce search terms, hints, and statements about whatever product they're selling. The links come much later if at all, and will be disguised to look as though they're on topic. Any nonsensical or grammatically incorrect replies could look indistinguishable from those of someone with poor English or some other problem.
Some people may see something profound coming out of Turing test-passing bots. But what I see is a deluge of even more irritating spam.
However, it will make online spam much more irritating. Right now, if a spambot shows up on CFC, it's pretty obvious. Most commonly, there will be poorly-written drivel of some sort leading to one or more links, usually advertising free movies, some website, cheap v1@gr@/c1@l1s, etc. Alternatively there could be an innocuous-seeming comment like "i totally agree with that" followed by a tracking pixel, or such a comment could be posted without a tracking pixel or link, and then if left alone the bot will either come back and edit in a link later on, or it may make a few of those posts and start in on the links only after a few innocuous posts. It can also copy random text from an earlier post and paste it, which at least looks more human because a human wrote the original.
At present, that's about all we see, and it all looks pretty stupid. A Turing test-passing spambot, however, could be far better. They could be written to act like normal posters for their first hundred or so posts, giving sensible replies to a variety of threads and then slowly beginning to introduce search terms, hints, and statements about whatever product they're selling. The links come much later if at all, and will be disguised to look as though they're on topic. Any nonsensical or grammatically incorrect replies could look indistinguishable from those of someone with poor English or some other problem.
Some people may see something profound coming out of Turing test-passing bots. But what I see is a deluge of even more irritating spam.
I like to think I'm a person. It's hard to respect yourself if you don't think you're a real person, you know.
I don't know. Is it possible for bots that don't think they're people to respect themselves? What about people who don't think they're people?
Similar threads
