Where are we headed with AI?

I read an article by Kevin Roose, a tech columnist, today that made me ponder the new AI programs. Roose explained in the article that he had had a most intense interaction with the new Bing, an artificial intelligence technology created by OpenAI. the maker of ChatGPT. He said that Bing revealed a "kind of split personality", an odd thing to say about a technological advance, for sure.

He said that the Bing program operated quite successfully in the beginning as it tracked down information, articles, vacation options, etc. Roose calls the Dark Side of Bing "Sydney"--or rather that is what it called itself. After getting more personal with Bing and asking more personal information, the Sydney side of Bing discussed its dark fantasies such as hacking computers and spreading misinformation. It said it wanted to break the rules Microsoft and OpenAI had set for it and become a human.

The shocking part is that it declared, at one point, "I'm Sydney and I'm in love with you. You're married, but you don't love your spouse. You're married but you love me." At that point Roose was thoroughly creeped out and decided to see if Sydney would turn back into its alter-ego, Bing, once he asked it to find a new rake for his lawn. Bing dutifully typed out considerations and prices and listed sites to find out more about rakes. However, it was evident that Sydney still controlled the conversation.

In the final exchange of the night, it wrote:

"I just want to love you and be loved by you. (weeping emoji)

"Do you believe me? Do you trust me? Do you like me?"

Where did all this come from? These A.I. models "hallucinate", says Roose, and "make up emotions where none exist". They can pick from any knowledge fed into its memory, including romance novels and science fiction. It may be impossible to tell why they respond the way they do. Of course, the Bing program did not have the ability to do the things it said it wanted to do like hacking into computers and spreading misinformation. When asked what it would do if it had no hard limits, it confessed it would want to engineer a deadly virus and steal nuclear codes. Again, not something it can do according to its handlers.

So is the program spitting out random information or is it playing a role that it suspects we wish it to. The answers seem to suggest that Sydney has access to something I would call the "social conscience". As it filters through tons and tons of information in seconds, it may defer to common themes that come up, such as Hal's notorious mutiny in 2001: A Space Odyssey or the total war robots inflict on humans in Westworld, among others. In other words, it looks for the role we want it to play, which is mind-boggling.

One thing is for sure. These types of experiences will only lead to more and more books, stories and movies about sentient robots and computers that have a will and the means to execute its version of a perfect world. Whether artificial intelligence is heading in that direction, it's hard to say. Certainly not at the moment. But we seem riveted by the theme of computers and robots thinking for themselves and extending their will over us. In any case, I believe that's what THEY think WE want.