“The most interesting thing from the research,” said Kevin Cornish, the director of Fall in Love VR, “is this premise that the thing that creates human bonds is not the words we say to each other, but the act of conversation.”
In Cornish’s new virtual reality project, released today for the Oculus Rift, users confront the question of whether it’s possible to experience intimacy with an avatar by sitting across from one of five photo-realistic actors and, one by one, asking many of Aron’s questions off prompt cards. Out loud.
The speaking-out-loud bit is key, as the potential love interests, looking adorable, yet vulnerable, respond only when the specific questions are asked. Ask or say anything else and they just sit there looking expectant.
That’s because Fall in Love VR, from Tool of North America, uses natural language processing –becoming among the first to utilize the technology in an interactive VR project–to make users feel like they’re truly having an intimate conversation. Cornish said he got the idea when working on a VR film with Taylor Swift. “There was one moment where [Swift] looks into the camera,” he recalls, “and it feels like she’s looking at you and talking to you. There’s a connection that you can get in VR and not any other medium.”
Added Cornish, “The idea is taking all the advancement in natural language processing and pairing that with an intimate conversational experience to give a sense of what it’s going to be like when we’re having conversations in virtual reality. It’s like that moment in [the film] Her, when there’s that question of how many people are you talking to, and how scalable is it [to have an AI say the words and have them repeated again and again to other users]….I only have to have that conversation once. It’s kind of like the VR equivalent of what CC meant for email.”
In short, the idea behind Fall in Love VR was to give users a conversational experience where the joy comes from the simple act of having the conversation.
. . . .
Although that gives the initial impression that this will be a two-way conversation, it really isn’t. The entire experience is built around you asking the avatar questions, and them answering. In the early stages of production, Cornish explained, the idea had been that the avatars would ask you questions as well, but that was quickly rejected because in testing, Tool found it put people on the spot, which left them feeling uncomfortable. The decision was made to limit the functionality to asking questions of the avatars and having them respond. So those interactions have to be as realistic as possible.
. . . .
“So much of a personality is based on a face,” he said, “It’s that idea of pairing natural-language and machine learning with the personality and the warmth and eye contact that come with having a photo-realistic human face.”
. . . .
Cornish is also fond of one bit of feedback he’s heard on multiple occasions: “We hear, ‘my wife would be jealous of this,’ or ‘my boyfriend would be jealous.’ It’s such an interesting thing in making a film that that’s the reaction….It really comes from the eye contact and the naturalness of” the interaction.