Humans and AI working together in a new sort of team.
Would you like to learn a language from a friendly piece of software?
Berlitz has been in the business of language instruction for 143 years. They’ve been looking to move forward with 21st Century technology, but at the same time “Berlitz has built our methodology and brand on delivering the best outcomes for students serious about fluency, which requires a very human centric experience,” said Curt Uehlein, Berlitz’s CEO.
Digital online resources for learning language are surprisingly sparse. Search for language instruction videos on YouTube and you’ll find a handful of short, often homemade, videos covering some basic vocabulary. Khan Academy, one of the leading sources for instructional videos, has little to offer at all.
Part of the challenge is scale. Berlitz attempted to record instructional videos with live instructors in a studio, but these were expensive and time-consuming to produce. Cranking out thousands of videos across a range of several languages was judged unsustainable. It’s a long time to keep Teacher Pat in the studio, especially if you want Teacher Pat to be as fresh and engaging in Video #997 as in Video #1.
Their solution is a company called Hour One, and AI company that “seeks to scale the human touch.”
In a Zoom interview, Hour One’s Natalie Monbiot explained the company’s process. They capture the likenesses of real humans, who are paid for their appearance. Those are then rendered as “characters,” fully animated faces that are on the right side of the uncanny valley.
MORE FOR YOU
Rendering text into talk is easy (about thirty seconds after this column is posted, you’ll be able to have it read to you by friendly software). Hour One’s trick is to pair that text reading capability with a friendly human face.
You can see in the accompanying video that the “characters” are convincingly real, as is the speech rendering.
While promotion for the partnership highlights the AI element, that element is only involved in creating the video imaging. The text that is rendered is written by the humans at Berlitz, so that students will not have to deal with AI-generated at content or lessons, a trick AI has not really mastered yet. The videos and the “characters” themselves live on the cloud, allowing simple access.
It’s a new sort of team. The AI-generated images are based on live humans, and they deliver lessons written by human experts. But it’s the software that manages the actual delivery of the lesson.
It remains to be seen if this sort of human/AI hybrid is an adequate replacement for human instruction. It’s a little disconcerting to see human teachers replaced by AI-generated characters, but the lessons themselves are still human-created, and when it comes to cranking out hundreds of language videos, we may be talking about a job that no humans actually want. The AI-generated presentation is certainly cheaper and easier to scale up than a human instructional video; Berlitz will have to wait to see if the students can stay excited and engaged by Cyber Teacher Pat through hundreds of lessons.