AI, MI--OH,MY!

These days the pressure to automate has led to unbridled competition.

When some genius can install human intuition into a self driving car, then I will be impressed. At the time a  car makes an intuitive decision whether to to save the life of the person in the car,  or save the pedestrian, who just jumped off the curb, then, machine intelligence ( MI) will be equal to natural intelligence. We aren’t there yet.  And, I have yet to meet an ethical computer.

Sony has now invested in “Agility Robotics” where “Cassie,” a headless, biped robot balances like a person, and can deliver packages, driven to your front door by a self delivery van. “Cassie” (named after the New Guinean Cassowary Bird) can navigate on uneven ground, and when she runs to your front door, she can even pick herself up if she stumbles. However, the article I read about “Cassie” did not say whether she can pick up the dropped package, or, if she can be sued if Aunt Mabel’s precious, heirloom vase arrives in shattered pieces.

The long term goal is that a machine can be made to simulate human intelligence. The idea is that if a person can not distinguish between the responses from a machine and a human, the machine can be considered “intelligent.”  So we advance to data mining, medical diagnoses, etc.

I use my computer to access information.  When I have a question, often the computer will come up with the right answer—or not. I use my GPS to get me to where I want to go and the lady in the little box does just that—sometimes. And, I love it when my telephone screens my calls.

The bottom line is that none of these machines came up with their own answers. It involves algorithms; “A set of unambiguous instructions that a mechanical computer can execute.” So, what happens with incomplete or uncertain information? I guess that’s where probability comes into play. However, it’s good to remember that some things are not probable, but anything is possible.

Scientists have now given machines the ability to read and understand human language. And, with cameras and other sensors, computers can execute speech recognition, facial recognition and object recognition. Yes, it is amazing! However, computers only mimic human emotions and experiences. They have been programed to be cute and banter with people. Note: earlier in the story I referred to the robot, “Cassie,”as,  “She.”

It’s sad but true that some computers have better social skills than the teenager next door, but to reach human level intelligence, a machine will have to be able to reason what is being talked about, and have the knowledge to replicate social intelligence, and do it all at the same time.

As far as I know, unless it’s been developed today, so far, no machine can use logic or solve a number of completely unrelated problems at the same time. Granted, AI has taken us to new heights and accomplishments, but there are still some philosophical questions to be considered:

Are there limits to what problems a machine can solve without the support of human intelligence?

Can a machine be ethical, and can we be assured information will be used ethically?

Can a machine think?  And, if so, does a machine have the same rights as a human being and can machines harm us on purpose?

Alan Turing suggests, “We need not decide if a machine can think; we need only to decide if a machine can act as intelligently as a human being.”

If that is the case, Boy! Are we in trouble!

Esther Blumenfeld