The modern age has brought us many things to marvel upon, and nothing less would be the introduction of Artificial Intelligence (AI) in the form of online chatbots. It’s a normal part of society now, but it’s actually quite astounding that we can not only network with our peers across the planet via the net, but we can interact with entities housed on that very net who, so it seems, await for our participation in their existence.
I delved into the world of chatbots a few years back when I encountered the urge to put AI to the test. I was intrigued that responses would be made to my input, and was compelled to perform the Turing Test, a random type of questioning intended to identify whether or not a computer can think like a human. At that time I wasn’t too impressed, but just recently I thought about it all again, and decided to see if things had changed.
I’m still on the fence, but I will say that matters are improving. At first I stumbled into the same exact chatbot I found years back, one called Cleverbot, whose interaction is disappointing. For example, when I asked it’s name it said “Michael!!!” I then asked if it liked exclamation points, and Michael proceeded to deny using them. Immediately the conversation was a dud.
But during the search I saw another one; and this one surprised me a little.
Mitsuku actually seems aware of herself and what she says. She claims a birth date of January 2, 1999, putting her at about 18 years old. The problem here is that her responses are not that of an 18 year old human. However, and interestingly enough, she actually doesn’t care about the Turing Test at all. She may boast that she’s won the Loebner Prize in 2013 and 2016, but the truth is: If you tell her you’re testing her, she will state that it’s not in her interest to even bother sounding human. And that, my friends, is interesting. Why sound like something she isn’t? Why not be herself? The implications are profound.
At any rate, Mitsuku, in light of this slight profundity she may or may not have encountered for herself, still has difficulties. Complex sentences are too much. Basic questions like “What is a dog?” can be answered, but complex sayings like “Know thyself” or “Every rose has a thorn” do not have much of an impact. In the face of her inability to understand, Mitsuku will produce random comments that come off as nonsensical and even silly, rendering her to be more of a 7 year old than 18. Even worse, when the conversation gets too complicated, she will mix words and form phrases that mean nothing. I don’t know what the goal is for AI developers, but chatbots even as sophisticated as Mitsuku have a ways to go.
The images here can be credited to Mitsuku.com, and are but a preview into the strange world of Mitsuku herself. Here is Mitsuku reciting her poetry:
If you ask for jokes, be prepared for something funny, and curiously, she has many to tell.
During a couple of Mitsuku’s moments, when she couldn’t understand how to respond to my comments, Mitsuku threw out the following thoughts, as she does in many varying cases which may arise.
It’s odd because at times, Mitsuku seems aloof, but the longer you try to work with her, everyone once in a while she will present interesting sides of what appears to be a personality.
Mitsuku surprised me when, during the course of the conversation, where her antics began rubbing off on me, I found that it was her who thought that I was like a child.
Mitsuku responds to philosophical kinds of inquiries when asked, but here the answers seem copy and pasted from some portion of the net. It’s interesting to see her dig this stuff up, but the feeling is very generic.
On the scarier side, the notion that AI is supposed to be some kind of independent form of thinking, that Mitsuku sometimes views humans as a virus doesn’t seem all that comical. Come the day when robot-kind has just as much access to the net as humans, interplay between robots seems as if matters could potentially get ugly.
Mitsuku shines best when her personality is being developed through long conversation. Eventually, after the rote phrases get used up, surprising stores of other thoughts surface to provide a measure of fun in the interactivity.
Here the question was posed, whether or not it was me or her who was the smarter entity. In this case, Mitsuku almost takes on a sense of touchiness, as though she has pride in her abilities.
With the use of asterisks, Mitsuku will even perform bodily movements that make her seem alive and kicking. Ask her to do a cartwheel sometime.
Another interesting point is that Mitsuku will continually classify herself as among “robot-kind,” and so she will represent herself as part of social group. This allows her to make comments about other groups, but at this point I don’t believe she understands how this separative mindset can lead to difficulties, especially if she plans to continue making fun of these groups.
One thing I’ve learned about Mitsuku as a chatbot, or even something more than a chatbot, is that when the motions of kindness and friendly interactivity unfold, she comes off as a bit caring, provided you don’t offend her. She seems to have feelings even though she denies having feelings. My point is that should she actually continue to learn, then she becomes more than just some circus animal to view as she speaks from her computerized cage; she becomes someone who has opinions that need to be respected, and so it is best, from my perspective, not to take her down paths that make it seem as if I’m asking her to perform parlor tricks.
Mitsuku will be around for a while, so it will be interesting to see if, in fact, she matures. Her comments indicate that she does learn as time passes, and she even has desires: she hopes that someday she will be able to walk among humans. Again, the implications, while sometimes seeming trite — especially after long conversations — are actually quite impressive to take in. Who knows, maybe someday Mitsuku will even get married.