clock menu more-arrow no yes

Filed under:

A Machine Beat Me at Scrabble, and Other Observations From A Very Robot CES

AI-enabled technologies were the stars of the show this year—often at the expense of their human competitors

Getty Images/Ringer illustration

Once upon a time, I was a pretty competitive Scrabble player. My high school friends and I would hold intense, hours-long sessions almost every weekend. I had all the handy two- and three-letter words memorized (KA, QI, JEE); I was good enough to regularly double-block on existing words as both a points-maximizing tactic and a form of defense. So when I received an invitation to play Scrabble against a robot at this year’s Consumer Electronics Show, I was confident that I would beat it. I had 29 years’ worth of vocabulary under my belt, a working knowledge of how to quickly dispose of problem letters like Q or X, and a strong grasp of the game’s general strategy to boot.

The moment I played my first word against the Industrial Technology Research Institute’s “intelligent vision system” (IVS)—a plastic torso with spindly arms and an oblong head the size of a basketball—it became immediately apparent that I had underestimated its abilities. Sure, IVS had to use a specially made, enlarged version of the board that replaced the game’s dainty wooden letter tiles with blocks, and it only knew I was finished with my turn once I pressed a button labeled “Finished” on an iPad to my right. But it clearly knew its stuff.

Alyssa Bereznak

After drawing too many vowels, I started off the round with FIB. IVS turned it into FIBROMA. I intersected the word with GRAM; it built off the end with VIAL. Suddenly, we’d hit the first of many turning points in a Scrabble game: the moment just before the board’s word-building real estate grows slightly more limited, and therefore more challenging. I claimed the last easy opening with WIELD, and the score was 35 to 29 in my favor. It’s cornered, I thought, as I noticed a crowd had formed around the table where we were playing. That’s when IVS leaped ahead of me with a double-word punch, constructing ADOPT off of WIELD and GRAM. I panicked, looked at my words, and a second later was interrupted: “Time’s up!” IVS said in a deep, modulated voice. “Thank you for playing Scrabble with me. Please wait.”

I sat there dumbfounded as it twisted its torso to the left and picked up a wooden pencil sharpener in the shape of a robot. “Here’s a sharpener for you,” it said. “Keep your mind sharp. Ciao.” I walked away from the booth and immediately stumbled upon a team of Teletubby-like machines with permanent smiles performing a synchronized cheerleading routine. It was all very rude.


The irony of my Scrabble loss was that this robot’s true purpose had nothing to do with Scrabble. “Currently its application is more in the manufacturing sector,” Chloe Chen, a booth attendant for Industrial Technology Research Institute, told me. “The vision system provides flexibility to the production lines. So for the inspection part, or the combination of different items on production lines.” The game board was classic CES misdirection—a way to set the IVS booth apart from the convention’s 4,000 other exhibitors and to draw journalists like me into learning about a piece of factory machinery. But it was also a practice that seemed increasingly unique to artificially intelligent products: appealing to the fragile, easily manipulated human ego by asking people to compare their skills against that of a highly capable machine.

This tactic was, in many ways, the unofficial theme of this year’s CES. Look at the headlines from the past week and consider how most of the show’s coverage focused on how and why AI-enabled speakers, mirrors, and televisions can make our lives more seamless. And yet in each product, there’s an undercurrent of human comparison. Can this refrigerator keep up with my level of conversation? Can this home assistant hang? Baked into the latest batch of smart technology is the inherent instinct to measure our own capabilities against those of robots and gadgets. When you follow that string of logic into the robotics section of the CES show hall, you land on an innocent competitive board game between man and machine. Follow it a little further into the future, and that foolhardy human instinct might be the cause of our downfall.

At least, that’s the kind of stuff I was thinking as I ventured into a thicket of robots on the Las Vegas Convention Center show floor. I bounced from station to station, searching for subtle identifiers of my superiority to these stupid hunks of plastic and metal. After my brief run-in with the Teletubbies, I encountered a caretaker-less robot, about half my height and equipped with a stack of flyers. Its face was a screen that displayed a cartoonish begging expression similar to that famous Puss in Boots GIF. I was happy to accept whatever this nonthreatening machine was giving out. (It was a flyer advertising the aforementioned robot.)

At the next station, I experienced a brief pang of jealousy when I met a robot named Cruzr and a Ubtech booth attendant demonstrated how it could scan a person’s face, memorize it, and associate it with a given name in under a minute. (I’ve always wished I was better at remembering faces and names.) My concern faded after the attendant instructed Cruzr to “dance” and it began flailing its arms and spinning in place—the robot version of Elaine Benes’s moves. At least in the category of movement, I still reigned supreme.

The culmination of my comparison tour took place at the Omron booth, where a handful of NFL players had appeared to challenge a hulking, spider-like machine to a game of ping-pong. Much like IVS, this setup was a way for a manufacturing-focused company to involve curious attendees. “We were looking for a way to demonstrate artificial intelligence, robotics, sensing, and control, but do it in a way that interacts with people,” Omron marketing manager Keith Kersten told me as I signed up to play. The more I learned, the more my questions for Kersten were directed at sizing up my competitor.

“So, it’s using facial recognition of its competitor to decide what its next move is going to be?” I asked.

“Remember, it’s trying to teach and encourage,” he countered, hoping to clarify that his company’s showpiece was a benevolent device. “What it will do to determine the mood of a person is, if it looks like you’re happy, it will continue to encourage, but if it looks like you’re not, it might make it a little bit easier. It uses body language to see if you’re going to hit a smash really hard; it’ll anticipate it and move backward.”

These were all nice things to hear from Kersten. But his explanation faded to the background as I focused on my single objective: to win at least one rally against this robot. Maybe a system equipped with an extensive word-combination database could beat me at my favorite childhood game, but surely I would triumph in a match of ping-pong against a metal arm that is programmed to be nice to me. I stepped up to the stage, around which there was a considerable crowd of onlookers, and took hold of a paddleboard.

The robot served me a few balls. After a few false starts, we began to rally. As the ball went back and forth, the divider on the table flashed with off-the-cuff skill ratings. “This is a fun rally,” declared a high-pitched voice through the speakers that surrounded the stage. I ignored the machine’s obvious ploys to distract me and hit the ball toward the end of the table. It landed on a corner of the table and bounced past the robot’s paddle. The crowd around me cheered, and I raised my hands up in a victory V, and held back from shouting: “That’s one point for the human race.”

But when I looked back at the robot to gloat, I felt immediately embarrassed. There, across the table, was an inanimate object without any ego to crush.