Despite feeling lost in translation, I've gotten to know the little guy pretty well. When I need to know the weather, I flash the little yellow RFID tag in front of it, and it rattles off the meteorological statistics of the day. Something went wrong with my Twitter feed, and I lost confidence that the rabbit could adequately represent me on my social media networks without messing something up in an embarrassing public forum. I’m reluctant to rabbit-tweet again. The webcam offers some fun moments of voyeurism into my kitchen, and the remote voice messages allow me to try to torment my dog, but none of this seems to be building a strong emotional bond.

Among the 91 comments on “Our Talking, Walking Objects,” an article I published in the New York Times Sunday Review last year, many expressed a vitriol that I hadn’t anticipated. “The vision of people getting emotionally attac[h]ed to a talking vacuum or their electronic diet coach is pathetic and sickening,” one reader wrote. Another responded, “I really want to believe this article is satire. I fear it is not. What a dystopian picture the author paints!” One active Times member eloquently quipped:

A robot your pet and your friend?
Have humans come to a dead end?
Empathy, compassion
Are now out of fashion,
A sad and an ominous trend.

In their dismay over our emotional machines they were validating their existence. Ultimately, the final few comments came from readers who championed the need for more camaraderie with our products, particularly in the area of health care and devices to assist the disabled. But the Karotz experience left me feeling like a phony. If even I couldn't feel any mojo between me and the rabbit, perhaps my predictions and assumptions were fundamentally flawed.

So after my disappointing Karotz experience, I went back to Georgia Tech to visit Dr. Andrea Thomaz at the Socially Intelligent Machines Lab and see the robot that put all these ideas of affection and social engagement into my head in the first place. Simon, the robot I had helped create, was joined by a new friend, named Curi whose shell I had also designed. Again, I was captivated by the subtle, coy head nods that told me Curi had seen me. I held up a toy ball and said "Please, take this," and the robot understood right away, lifting her arm to offer me a hand so she could clasp the object. She held it up to her eyes and shrugged her shoulders to admit, "I don't know where this goes." (The robot is programmed to sort toys into various bins by color, and lets you know when it encounters a color it hasn't seen before.) Even in a moment of ignorance, Curi was completely enchanting. She spoke to me in a human way; she made socially appropriate gestures; she anticipated what I wanted to do. The exchange was so natural that I was able to suspend disbelief long enough to temporarily forget that I was interacting with a machine. I could simply ask it what I wanted to do in an intuitive, human way. My faith in the potential for smart objects to provide helpful assistance while also making an emotional connection was restored.

So what was it about Karotz that fell short? The first aspect is trust. Since my Karotz loses its connection to the Internet on a regular basis (perhaps through no fault of its own), it takes a long time to boot up and makes it hard to know if it is working. The second is clear communication. With its cryptic expressions and vague feedback, the device difficult to understand. The commands it recognizes are very particular (e.g. “Karotz weather information,” “about time,” and “movie config”). The third (and most challenging) failing is Karotz’s inability to learn my preferences and grow more sophisticated over time. What if it knew my calendar and could tell me to take a raincoat before I left on a trip? What if it could hear from my tone of voice that I was stressed and provide a diversion to relax me? What if it remembered what genre of music I liked to listen to when I need to unwind at the end of the day? These characteristics would require more complex programming and more dependable hardware, but they are not beyond the realm of possibility.

The tools for meaningful digital-physical integration are finally accessible, but it’s still a messy challenge to get them all to work together in a meaningful way. Dreaming about robots is a bit like dreaming about finding strangers who will understand you completely upon first meeting. With the right predisposition, the appropriate context for a social exchange, and enough key info to grab onto, you and a stranger can hit it off right away, but without those things, the experience can be downright awful. Since we’ve got a lot more to understand when it comes to programming engagement and understanding, the robot of my dreams is unlikely to be commercially available any time soon, but with the right tools and data we can come pretty close.

An ongoing series about the hidden lives of ordinary things