Why the Old-Fashioned Dream of Humanoid Robots Refuses to Die
SoftBank put its robot Pepper on hold, but replacements are in the wings
When SoftBank froze production of its humanoid robot Pepper recently, professional robot watchers started writing obits and smirky I-told-you-so’s.
SoftBank has since taken pains to say it is only reorganizing a bit, but sales figures after seven years on the market show it hasn’t been easy: Only 27,000 units sold worldwide. For comparison, Sony reportedly sold more than 11,000 of its Aibo robot dog in Japan in the first three months of 2018 alone.
Pepper’s struggles matter because the vision that drove its development persists among roboticists today.
And what is that vision? Pepper’s original creators, Aldebaran, planned for it to be the first humanoid “companion” robot for consumers to take home, live with, and (hopefully) love.
Its creators equipped Pepper with AI so it could read and respond to human emotion via speech and gesture. They gave Pepper expressive hands and giant eyes for those purposes and sought to endow it with the ability to carry on a conversation. To understand the scale of that ambition, consider that even today, eight years later, Alexa, Siri, and Google Nest still struggle with follow up questions.
The ability to make conversation also introduced the fraught questions of gender and voice. Executives chose to imbue Pepper with the wit and curiosity of a twelve-year-old boy, presumably to give it license to ask a lot of questions to support machine learning.
But “companion robot” is not quite the value offer to average consumers that it might seem at first glance. I say this having been a consultant on the U.S. team at Sprint that was developing marketing plans to launch Pepper in 2013. (Technical issues delayed actual launch until 2015.) Reviews in the press in the years since launch echo what we heard in testing.
Even among very tech-forward, the first request of a robot is that it be useful. People asked some version of “Will it wash my dishes?” in nearly all the groups we spoke with. That’s not unreasonable given that Pepper had a price point of almost $2000 and monthly data fees in the low hundreds. But even with jointed fingers and an opposing thumb, Pepper was not equipped to so much as bring you a drink.
Its humanoid form is a surprising sticking point. Roboticists say that raises expectations too high, leading to disappointment. Others have pointed out that your cell phone can do nearly everything Pepper supposedly can, except follow you around the house. And even there, Pepper has its limits. It can’t climb stairs, because it moves on wheels.
Perhaps the hard-plastic, friendly humanoid is just an old-fashioned vision of the future.
So what, then, keeps the dream alive?
Most likely it is that there is a market for something of this nature that is too big to ignore.
Where SoftBank once envisioned selling Pepper to consumers, it has since shifted its focus to industry. Its pitch to categories like retail, finance, and healthcare on its website seems built on two main pillars: Novelty, and the ability to tolerate repetition.
Novelty only works for first encounters, the “one day of fun” dreaded by toy makers and parents alike. But a tolerance for repetition is the same virtue that gives robots entrée into factory work.
That would seem to make it a natural for certain roles in healthcare. Who better to monitor and endure the repetitive conversations of Alzheimer’s patients than a robot? Estimates say roughly 44 million people worldwide are afflicted with Alzheimer’s. Another 75 million are “on the spectrum” for autism. Looks like a market exists.
Even if SoftBank does decide to pull the very literal plug on Pepper, other robots vie to take its place. One stands out for sheer similarity. Lurking backstage like the aspiring Harrington in All About Eve is Abel, yet another emotionally responsive robot modeled on a tween-age boy, this time with an expressive face, instead of giant eyes in hard white plastic.
For all the very substantial reasons to be wary of robots and AI, this suggests another, more prosaic warning: Fall in love, but don’t get too attached. Just like your laptop, you’ll probably want the upgrade when it comes out.
