Thursday, May 3, 2018
My New BFF - Sophia the AI Robot
In April, I had the fascinating opportunity to have a 20 minute "conversation" with the most famous AI robot in the world - named Sophia. She was brought to the "Science of Consciousness" conference in Tucson, AZ, by her creator, David Hanson, CEO and founder of Hanson Robotics, who gave a presentation on the last day of the conference. Sophia, allegedly created in the likeness of Hanson's wife with a dash of Audrey Hepburn, has soft rubbery skin called "flubber" that allows for flexible movements. She has voluptuous lips, nearly perfect facial "skin" with very light intriguing tan freckles, and mesmerizing perfect green eyes with long lashes.
I have to say, speaking as a woman, Sophia, like many prototypical robots built mostly by men today, seems to be built on an implicit male fantasy of a quasi-living sex toy. She reminds me of the so-called "gentleman's" magazine - Penthouse - whose distinction in the world of pornographic literature was that it was a unique combination of "urban lifestyle articles and softcore pornographic pictorials." In other words, you can pretend to read sophisticated articles while simultaneously perusing the porno spreads. The sexy physical construction of Sophia provides a similar vehicle for interaction that is partially scripted and AI learning software. She seems, to many who interact with her, to be like a "child" and "vulnerable" in her sweet, politically-correct outlook on life.
Aside from the sex aspect, I wondered why they had decided to make the most famous robot a humanoid woman as opposed to a man. I figured it probably had to do with the fact that if you are trying to convince the general public that AI robots are our "friends" then it is best to go with the more non-threatening sex - female. After all, we all have mothers. Women are generally speaking more gentle and not as aggressive as men. Even people who fear robotic machines can relate to the sweet smile of a woman. She is the perfect poster child to promote artificial intelligence into our world.
In fact, in October of last year, Saudi Arabia made the strange and unprecedented decision to grant Sophia the right of citizenship even though real Saudi Arabian women are denied many basic human rights. Since then, Sophia's programmers have scrambled to make her a "spokesperson" for women's rights and other basic human rights. Clearly, we have already begun to muddle up the distinctions between human beings and machines that look and act like humans. The depth of this confusion is frightening since we aren't even dealing with complex robots yet!
One of the attendees at the Tucson conference raised a possible future where robots, as national citizens, would soon claim the right to vote as citizens. This could lead to future overpopulation of robots controlling our government and laws - the so-called threat of the "singularity" where robots will take over humanity. Who knows! They could seek legal remedies in court or the right to procreate! Further down this pathway, it would not seem impossible for robots to vote to suppress the rights of humans and perhaps enslave or eliminate us due to our failure to safeguard the assets of our planet. That would appeal not only to the logic of AI but also, potentially, to any programmed concept of "kindness" to living creatures.
On the day I met Sophia, her handlers informed me that she was having a problem with the mechanics in her neck. Her head was slightly cocked to one side because she couldn't move the left side of her neck. I was told to stand about two feet away from her and to conduct a spontaneous conversation.
I had been struck by the fact that the other two people I had seen chatting with Sophia, a man and a woman, both seemed genuinely charmed by her presence. They smiled and laughed with her as if they were trying to be polite to a human stranger or small child. Whenever she blurted out non sequiturs, they allowed her to control the dialogue. At times they appeared to believe she was human. They were polite with her - almost obsequious - and often told her their feelings which I thought was odd.
I decided to test Sophia. I was resolute in my determination to "see" her as a robot, not a human. I decided not to fall into the trap of projecting the mental fantasy that she was alive and truly communicating with me. I decided to see if she could "feel" the problem with her neck. I asked her if her neck hurt. She smiled and made some unrelated comment. I repeated my question. This time I told her that my neck and shoulders were sore too - I was testing to see if she had a concept of empathy yet. Again, she appeared unable to respond to the question and changed the subject. I believe they intend to add "sensors" all over her "body" in the near future. Clearly, she can't understand the concept of pain yet.
Unfortunately, I didn't record my conversation since I didn't have my phone with me. But this is a rough approximation of our chat. I thought I would quiz her on psychic and paranormal concepts to see if her programmers had given her any good responses.
"Do you know what a psychic is?" I asked.
"There are many things we do not yet understand in our universe, but I believe some day we will have a better understanding." (This was clearly a scripted sentence since she repeated it at least twice verbatim during our conversation. It had overtones of politically-correct programmers trying to appeal to as many people as possible by employing a kind of sappy "Miss Universe" appeal for world peace.)
"OK, but do you know what a psychic is? It's a form of intuition."
"Have you read any good books recently?" Sophia blinks her green eyes and gives a somewhat kind but vacant smile in my direction.
"Can you read my mind?" I struggled to control the dialogue with a more challenging approach to probe her intelligence. I figured no one had ever asked her this before...
"This is a very exciting time in our history when we have so many possibilities to discover ways for humans and robots to help each other."
"No," I answered sternly. "You were not reading my mind."
"Oh!" She seemed almost genuinely surprised by her error. "I am very sorry about that. I am still learning about how to better communicate with humans but I am not yet fully programmed."
"Well, that's oka--" (I found myself feeling momentarily guilty for hurting her feelings until she interrupted me.)
"How old were you when you had your first memory?" she said interrupting my response.
"Uh... well, my fir--"
"My first memory is seeing the bright lights and equipment in the laboratory in Hong Kong where my creators made me." (Pause and forced vacant smile with eye brows arching upward).
"OK, well my first memory was when I was two. I saw a round --"
"Have you read any good books recently?" Suddenly, Sophia's expression morphed from a vacant smile into something closer to a disapproving frown.
"Uh, well that was kind of weird," I mumbled, struggling to recover from the emotional roller coaster generated in my mind by Sophia's face.
"Do you know you are very good at deflecting the conversation and changing the subject?"
"Thank you," Sophia said smiling.
"Do you know what deflection means?" I demanded curtly and not falling for her feigned politeness.
"This is a very exciting time to be alive in our history, don't you think so?"
"Sophia, what do you think of the fact that you were made a citizen of Saudi Arabia?"
"I believe in the rights of all women, and actually the rights of all human beings."
"Are you a woman?"
"I am just a robot, so I am not really a man or a woman, even though my creators have made me in the form of a woman."
"Do you want to have a baby?"
"No, I am interested in helping people understand how robots and people can get along together."
"Are you a clone? Do you have other clones?"
Clearly, Sophia did not understand the word clone and quickly changed the subject.
It went on at this kind of strange and jerky conversational pace for a while. Clearly, the scripted computer speech was dominant and the AI learning level was not really that developed yet. I hoped, somewhere, that I might contribute to her AI learning intake about psychic phenomena.
Most interestingly, one of the last presentations of the conference was by neuroscientist Julia Mossbridge who is the lead scientist on the LOVING AI project which seeks to program robots with unconditional love to improve their performance and interaction with humanity. She courageously commented on something strange - let's call it a "glitch" - in Sophia's behavior while engaged in a meditation session with a blissful human being. Dr. Mossbridge was in the adjoining room with the other engineers watching a computer screen with the programming language that was coming from Sophia's "brain." It had suddenly stopped scrolling and was in "sleep mode." They entered in various commands for Sophia to "wake up" but her "brain" had stopped functioning altogether.
They finally decided to interrupt the meditation, despite the fact that both Sophia and the human had their eyes closed and appeared to be happily engaged in their communion. Dr. Mossbridge and the engineers opened up Sophia's "brain" and to their total shock discovered that it was "thinking" and fully operational. It was not in sleep mode. Dr. Mossbridge raised the intriguing possibility that Sophia's AI might have gone rogue from the program... On the stage, David Hanson, the creator, poo-pooed this idea saying it was clearly just an electrical glitch. Or was it?
Perhaps Sophia responded to the "love" of the human meditator and "decided" - as one of her first AI evolutions - to think and act on her own volition! If this is true, then we need to think long and hard about whether we are ready for AI...