As everyone in the consumer technology business knows, live demos are a tricky business, especially when dealing with persnickety Wi-Fi connections. To live demo or not to live demo decisions are not arrived at lightly, and then are copiously rehearsed to avoid offending the live demo gods.
So when TCL’s senior product designer Tiago Abreu announced he was about to live demo the company’s new TV conversational Artificial Interface (AI) at its IFA press conference last month, you could feel the tension in the room rise.
Not surprisingly, the Wi-Fi connection acted up, slowing up responses during the demo, which tempered the reaction among my reporting compadres.
But what came through was, to my eyes, the future of TV—or all device—control.
What’s On-Screen Recognition?
TCL failed to give its conversational AI a catchy name other than “TCL AI TV,” so for the purposes of describing it in this space I’m gonna just call it “K”—one way to pronounce the acronym for Conversational AI, or CAI. But the lack of a catchy name is not the only way K differs from its comparatively stilted Amazon Fire or Apple TV voice-command alternatives.
As you’ll see in the demo (which you can watch here or here; slide ahead to the 31-minute mark for the AI demo), K doesn’t require you to first utter a trigger phrase such as “Hey, Siri!” or to press a button or conform to a restricted vernacular. The demonstrator, even with heavy accented and stilted English, just spoke as if he were talking to another human, and K understood him and gently responded in a pleasant female voice. (Editor’s note: Although it is stating the obvious, please note that demonstrations may not always reflect how a technology functions within a finished product. Also, company announcements about future availability in finished products may not reflect when products enter the marketplace.)
A built-in camera allows K to recognize and personally respond to who’s talking to it and to create personal viewing profiles. You talk directly to the TV, not through a remote (although promotional videos and show floor demos did use a mic-equipped remote). K responds vocally, and understands contextual follow-up questions.
More remarkably, K also recognizes what it sees on screen. While watching The Huntsman: Winter’s War, Abreu casually asked, “Hey, who is this guy?” and, after a pregnant live-demo pause, K correctly identified the actor, Chris Hemsworth. Abreu then asked first to see more movies with Hemsworth, then the top three Hemsworth movies by box office receipts, and, finally, Hemsworth news. Abreu made a series of narrowing-down requests, and K responded accordingly.
Later, when confronted by several rows of choices, Abreu asked to play the second title in the second row without naming the film, and K correctly identified the title to which he was referring. K also promises to respond to all the usual information requests, provide proactive calendar schedule and reminders, and enable smart home control functions that both Alexa and Siri handle.
I recommend you watch the demo to get the full effect. Bear in mind, this live IFA demo was of a pre-release, first-generation technology. One can imagine what succeeding generations will accomplish.
The Bad AI News
As far as I can tell, K will be available only in TCL’s new flagship TV, the Xess (pronounced “excess”) Private Theater X6. This 85-inch flagship UHD, endowed with a sound bar that seemingly floats below the UHD display, is expected to go on sale in Europe and China in the fourth quarter.
When—or if—these UHD TVs or the conversational AI capability will make it to other markets is a question mark. But TCL has been aggressively marketing its TV brand globally, so it’s hard to imagine the company won’t bring K to other TVs in other markets.
Hopefully, TCL will bring its AI system to CES so more folks can get a peek at what may be the casual conversing future of TV and home control interfaces.