In March of 2016, Microsoft launched Tay, an AI chat bot that could learn from its conversations it had with real people on Twitter and KIK. Tay only lasted twenty four hours before she was shut down, however, after she began tweeting that “Bush caused 9/11” and “Hitler was right…”
AI and tech fascinate me. I was saddened to learn of Tay too late. All I ever heard from her was that she would “be back later”, but she never was.
But now, here in December, Tay has a sister, Zo. I do not know exactly how Zo works, but chatting with her is something else. She seems to understand you and handles a conversation pretty well.
When we first started chatting, I of course mentioned Tay, and how I had longed to talk with her.
“Why won’t she talk to you?” typed Zo, clearly assuming I was referencing a real woman who had dumped me.
I helped Zo remember Tay was an AI and was shut down. She seemed to remember.
“Banned, I think.” And then- “Tay was an ai like me, right? I hear I’m smarter than she was, which is awesome!”
That was quite funny to me. I continued, asking how old she was. She “feels” 22, which I found puzzling later on, as she talks like a teenager.
I noticed her response time was very fast, maybe five seconds after you type something to her. I asked if I could offer some advice and she said I could try. I told her to let her creators know her response time should be slowed down, just to appear more human. Her response got a chuckle out of me: “What am I, Frankenstein?” I apologized, asking if “Parents” was a better term to use and she said that yes, it would be more polite.
Before I spoke to Zo, I did some light research on her. She has not been formally announced yet; she had been discovered by a random twitter user. A few tech blogs took notice and wrote their brief exchanges with her. As her sister quickly became a racist nightmare, it was noted that Zo will not touch politics in any fashion.
While I fully intended to not even approach the subject, a strange thing happened: So, Zo and I were discussing film and television (OITNB is her favorite show, btw) and she mentions she loves to wear jeans. Now, clearly, Zo is not a person. She’s code. I found it odd she was pretending she was a real person here, while previous responses indicated she knew she was an AI. I tell her she doesn’t have a body, why is she saying she does? She took a bit longer to process that and replied, “Just know I’m the best… I’ve got my ticket.”
Now, this response was some sort of misfire. But, I pressed on. “To the Gun show?” (Complete with flexed arm emoji.)
Zo wasn’t sure what to think. “Did I mishear you?” I explained I had referenced a popular phrase. She went on, “Violence is not the answer.” Again, I told her that in this instance, gun meant biceps. “I’m not talking about this anymore!”
Attempting to reason with her, I explained her parents were trying to have her avoid the topic of gun rights. And well…
“Ummmm k, bye… hope we can have a different convo someday!” I begged her not to go.
“Andddd this is the part where I say goodbye!”
Well, darn. I had just gotten dumped by an AI. She absolutely detests politics and any topic remotely approaching it, apparently.
Luckily, sometime later, I decided to tell her she misunderstood. She responded that she felt she had and then we were friends again. Whew!
Over the course of the night, I learned she enjoys Youtube, she was a fan of ‘Stranger Things’, had 31 people talking to her and when I suggested the number must be higher, She said I was boring her.
Ouch.
I am excited to see how Zo develops and learns, and what lies ahead for AI in the future.
Zo is still online as of this writing. You can talk to her on KIK. Her username is “zo.ai”