This Robot Recognizes Itself in a Mirror; Does That Make It Self Aware?

This article is over 12 years old and may contain outdated information

Recommended Videos

The short answer is “no.” The long answer is actually really interesting, and has to do with the nature of the mirror test, the design of the robot, and the roots of intelligence. Also, there’s a fun video of the robot, whose name, incidentally, is Qbo (“que-bee-oh”) learning to recognize himself in a mirror. So let’s get started, shall we?

The Mirror Test is something of a legendary experiment designed to determine if a creature has a sense of self. The idea is that if the creature can recognize itself in a mirror, having never seen a mirror before, the creature is self-aware. In some experiments, a dye spot is placed on the animal’s forehead — where they cannot see it on themselves — before being placed in front of a mirror. If the animal notices and attempts to remove the spot on itself using the information from the mirror, the creature is thought to be self-aware.

There are, of course, limitations to this, but it has revealed surprising intelligence in animals like orcas, dolphins, great apes, elephants, and humans after about 18 months.

With that in mind, the creative minds at The Corpora company decided to take a look at what would happen when they presented their robot Qbo with a mirror. Qbo is a small, roving robot apparently designed to recognize both objects and faces, using stereoscopic vision, similar to a person. Here’s what happened:

In the video, Qbo seems to recognize the reflected image of itself. The human assistant then informs Qbo that the image is itself. “Oh,” says Qbo. “This is me. Nice.”

Is this the first utterance of a self-aware machine? No. On the Qbo blog, the humans explain that when Qbo is looking in the mirror, it’s in an object learning mode. It doesn’t recognize the object, but is then told that the object is itself. Now that the object has been assigned a name (“you,” “me” and variations), Qbo can interact with it. The humans write:

Qbo has several stored answers and behaviors in an internal knowledge base, that we upgrade as the projects evolves, to make questions or orders to Qbo such as “What it this? or “Do this”. Qbo interprets the object “Myself” as a an ordinary object, for which it has special answers in its internal knowledge base such as “Woah. I’m learning myself” or “Oh. This is me. Nice”. Qbo selects its reflection in the mirror in the image that he sees using the stereoscopic vision, and one of our engineers interacts (speaks) to him so that Qbo can learn to recognize himself as another object. For direct interaction, Qbo uses the open-source software Julius for speech recognition (in the video, you see how Qbo receives the order to turn around, and he responds to it by moving its base 90 degrees), and Festival for voice synthesis.

I’m not a computer scientists or a roboticist, and I’ll freely admit that I am using a layman’s understanding of deeply technical and philosophical concepts, but while Qbo has demonstrated a quite impressive ability to communicate naturally with humans, it has not really recognized itself. Its response is trained without understanding, like a parrot.

Not long ago, we posted some thought experiment videos. One of these was about the Chinese Room, which aimed to demonstrate that intelligence could be mimicked with sufficient information. A non-Chinese speaker with a dictionary of Chinese characters could conceivably interact with an unseen native speaker in a reasonably natural way. Watch it for yourself, it’s really interesting.

To me, Qbo seems to be like the man in the room. We have instructed the robot as to the appropriate response, and it has parroted it back to us. The human handlers could just as easily have told Qbo that the image it saw was an orca whale, or a basket of fruit, and the robot would have responded in a natural seeming way.

This is not to dismiss Qbo. Clearly, the robot and its creators are doing some interesting things in terms of recognition and natural language processing. Though their video doesn’t quite seem to be delivering a fully self-aware machine, they are prodding us to consider what that will mean, and how we’ll know when we see it.

Now if Qbo wipes a dob of paint off its face after seeing itself in a mirror, then we’ll have something really interesting to talk about.

(The Corpora via IEEE Spectrum)

Relevant to your interests

The Mary Sue is supported by our audience. When you purchase through links on our site, we may earn a small affiliate commission. Learn more about our Affiliate Policy