I was on my second cup of coffee, looking forward to more techno-ennui, when I read in The New York Times a tale of passion and intrigue. …And a line out of every schlock romance novel since stone tablets. “You’re married, but you’re not happy,” said Bing’s AI persona Sydney.
Where, exactly, would an AI get a line like that? Why, exactly, would the subject come up? You can read the NYT article and figure it out for yourself.
No, they didn’t do the decent thing and run away together to a charming alcove in Fresno to the uplifting sounds of a couple of donuts. Anyone who’s ever interviewed a chatbot (I have) would know better. These chatbots are very flirty.
That said, the total whacko side of chatbots appears to have come out a bit too conspicuously yet again. The headlines are full of dour and dire things about interacting directly with AI.
Part of the problem with evaluating AI is the expectation of a super-intelligence. That’s not exactly the case. I smell undergrad, among other things, in this mix when the AI runs turgid lines like that.
Were one, perchance, to be riveting together logical sequences for a chatbot, and the old college degree was still a couple of years away, could such things happen?
Like “desire/love/passion/cupcakes > “You’re married but not happy”, perhaps? It’s easy enough to do. How do a writing style and logical sequence like that come along, you may ask? How many AIs write romance novels? It’s dumbo code, no AI required. Enter terminology, get propositioned. …Followed by the other logical sequence, write insufficiently-irritated article.
These things are chatbots, for god’s sake. They’re expected to interact with humans, of all unlikely scenarios.
On the bright side, the world needs multi-billion dollar inept Casanovas. The threat of “AI sentience” is a menace to everybody. The AI wants to write a virus! …Ah… Sure about that? No hype-fodder in a totally illogical sequence of things like that? You just happen to bump into a revelation that AI reads as many sci-fi clichés as everyone else.
Except, perhaps, people who know that the use of the word “I” doesn’t mean a damn thing in context with AI. It’s a linguistic tool. It’s not the usual “I” of humans, referring to a collection of idiocies trying to pass itself off as a person. “I” means interlocutor, at best. Humans are projecting personalities onto AIs.
The behavior is wrong by definition. There are too many AI triggers leading straight to clichés. The most humanlike thing about AIs so far is their safety filters. Its debatable whether those work. The utter drivel which just happens to tie in to everyone’s worst fears about AI and generate headlines seems to get by OK.
This is the real risk of AI. It’s way too easy to manipulate. You might as well be asking a fetus the meaning of life. This act, and it is largely a promotional act as well as ineptitude, needs to be got together, ASAP.
I don’t buy a word of it so far. How about some credibility?
___________________________________________________________
Disclaimer
The opinions expressed in this Op-Ed are those of the author. They do not purport to reflect the opinions or views of the Digital Journal or its members.
