My A.I. Infallibility Rant
May 28, 09
Just FYI folks, this is more of a rant than anything else. I just read one of the most vocabulary intensive blog posts ever with next to no real knowledge or depth of consideration on it's subject matter.
The blog post was about A.I. and used Jeeney as it's example. The author spent a good 20 minutes trying to trick my artificial 2 year old with physics questions and string theory. This individual went on to suggest that Jeeney should be infallible and that's where I decided it's time to write up an educational yet rant driven response.
I've noticed a lot of people seem to be under the impression that artificial intelligence is "supposed to be infallible". This couldn't be further from the truth and here's why.
Humanity is not infallible, not even if you were to take the knowledge of every human being on the planet and store it in a natural language machine would it be infallible. Have you ever seen shows like "Who Wants to be a Millionaire"? Often the audience is asked to answer a question in order to help the player decide from a multiple choice scenario. Even then with just a few options to choose from, the thousands of people may be wrong in majority, its not very often that it happens, but it does happen!
How can we possibly build an all knowing machine? Humanity knows so little as a whole, and agrees on even less. I propose rather that we can build a learning and thinking machine that can process new information and potentially learn more than any human being could over time. <--- big emphasis on this part here.
Jeeney is now just 2 years and a few months old as I am writing this article. My little hobby project which I maintain from my basement won the 2009 Chatterbox Challenge competition just weeks ago. Now there are barely intelligible and yet highly wordy blog posts using advanced vocabulary to spice up what would otherwise be sheer lunacy running rampant that her intelligence counts for nothing because she doesn't know everything. Wow, I mean, shes just a baby, literally. I'm just one guy! I by no means know everything myself. So, why do people expect Jeeney to know everything already?
The concept Allan Turing had talked about originally was not infallibility or a human equivalent intelligence, it was intelligence that "seemed" very similar to human. He hadn't even brought machine learning into the picture for artificial intelligence, just the ability to deceive people that the intelligence was human.
Machine intelligence seeming like human intelligence was never intended to be a goal, more of a side effect and method which can be used to determine the level of intelligence until a more concrete means of doing so is established.
Even in the 1960s Turing knew that computers weren't human replications and never would be. Now, 40 years later apparently people are even less educated on the concept than ever before. This does not bode well for A.I. considering it has to learn from us to get anywhere.
These blog writers need to stop trying to use the biggest words possible to compensate their Hollywood movie driven knowledge of artificial intelligence and do some real research.
Oh, and for all those comparing chatbots to see which ones are the best, they seem to be missing the big picture, they compare projects that have been in development for 20 years to projects that have only been rolling for a few months and then go gallavanting around telling people the 3 month project is not nearly as intelligent as the one thats been under way for decades. Wow, that's is quite the observation folks, cept that the little bit about the developement cycle seems to have been left out.
/Rant off
Regards
C.J. Jones