Why A I Isn't And Currently Cannot Be.....

[ToraToraTora]

https://news.sky.com/story/mother-says-son-killed-himself-because-of-hypersexualised-and-frighteningly-realistic-ai-chatbot-in-new-lawsuit-13240210

Classic illustration of function without realisation. I have always hated the term AI anyway and this sort of thing illustrates why perfectly. Clearly it was programmed to encourage his endeavour but not programmed to understand what the endeavour was. Long way to go I think.

[Canary42]

I first looked into AI as a research project in the mid-80s.  I was not impressed, all marketing hype, nothing at all like intelligence.  Although things have moved on since then, I still think the term is misleading - in fact I wonder whether anything non-organic can ever be called intelligence. Applying the artificial prefix attempts to validate the term, but I can't help thinking it's wrong. Chatbots certainly aren't intelligent, they're just human-programmed responses to key words and phrases as far as I can see. The computer isn't doing any real thinking, it's responding to human-programmed stimuli. Just my view, for what it's worth (I admit I've not kept up to date with AI since my original project).

[ToraToraTora]

yes canary, what changes is ever more powerful processors and more and more data storage allowing programming to appear more "intelligent". I have always said it should be called psuedo intelligence.

This case illustrates the problem perfectly. The bot spotted a desire and encouraged it. The flaw is that the desire is recognised but not the content of said desire hence the enouragement was the same as if the lad had said something positive. It could have been better than that, the ability does exist, this was clearly a very sloppily programmed bot.

[tomus42]

Surely this is what everyone's worried about. AI isn't going anywhere ttt. Things can't be uninvented once they've been invented.

[ToraToraTora]

tomus: Nothing has been invented. Better processors and memory have allowed bigger more complex programming that's all. Things can now be programmed that previously could not because of hardware not because the techniques are unknown.

People are right to be worried though, this case illustrates the problems very well.

[tomus42]

Yes it's worrying, but as I said it isn't going away. We'd all better get used to living in a world where you don't know what's real and what isn't. 

The Turing test has already been passed. Think about it'll be like in 10, 20, 30 years time.

[ToraToraTora]

The Turing test has not been passed:

"No AI has yet successfully passed the Turing test, but some say that AI is getting closer. The Turing test is an experimental method that measures a machine's ability to simulate human behavior. The test is considered passed if a human can't tell if they're speaking to a machine or a human after five minutes of questioning. 

In 2014, a chatbot named Eugene Goostman appeared to pass the Turing test at the University of Reading. Goostman was programmed to simulate a 13-year-old Ukrainian boy, and 33% of judges thought it was human. However, some AI experts disputed the result, suggesting that the contest was biased in favor of the chatbot.  "

[tomus42]

//2014//

You're having a giraffe. I could be an AI chatbot for all you know.