In an echo of studies that say more than half of U.S. households will have a smart speaker in the coming years, the 2018 Tech Trends report released today at SXSW predicts that by 2021 more than half of all computing in developed nations will be performed with voice. To deal with this massive shift in computing, AI assistants will have to become more harmonious with people and intelligent about emotion, Google Empathy Lab founder and principal Danielle Krettek said.
“It’s basically time for technology’s EQ to equal its IQ,” Krettek said. “I think we have a lot of brute IQ, a lot of functional stuff that can be done, but this new paradigm is inherently social, and when you think of language, it’s like the marker of our species, it’s like we’re connecting with these things in a way where it can’t just be that functional cognitive layer. We’ve got to design for the emotional layer.”
Krettek shared her remarks Saturday at YouTube HQ at SXSW in Austin, Texas during a panel titled “The Power of Story Through the Lens of AI.” She shared the stage with director Oscar Sharp and Ross Goodwin, creators of Benjamin, an AI model trained with scripts from thousands of movies that in 2016 was used to create Sunspring, a movie script written by an algorithm.
When art is created by both humans and algorithms, questions can emerge about who deserves credit, Goodwin said, but people will have to change the way they think about art made by machines and humans.
“It’s still the human beings who wrote the thousands of screenplays that went into Benjamin’s algorithm,” Goodwin said. “There’s a cultural shift that needs to happen, as well as a technological one. The notion that you can connect with a gestalt form of humanity and that it can be a meaningful communication between you and that [machine] would be a significant cultural shift.”
Krettek agrees and thinks many AI projects have skipped steps in the evolutionary process in order to chase what she calls the false grail of the humanoid machine. Some of that intelligence, she said, should come from an understanding of basic human responses that are precursors to empathy, like the mimicry that happens when one person yawns and other people instinctively yawn. Technology changes constantly, but human biology is a constant.
Empathy from a machine isn’t actually possible, Krettek said, but her hope is that assistants can someday achieve a kind of “empathic leap” whereby an assistant with intelligence and a personality created by the Google Assistant team achieves a connection that makes it feel more like a copilot at your side. It may not be human, but it would be closer to you and more understanding than a sterile robot.
You’d be forgiven if you’ve never heard of Google’s Empathy Lab before.
Krettek told VentureBeat that she has no background in machine learning or computer science, but the Empathy Lab has quietly advised the Google Assistant team and Google Cloud chief scientist for AI Fei-Fei Li for two years.
The daughter of a surgeon and an artist, Krettek told VentureBeat it’s her job to combine elements of art and science to provide a “framework for human empathy” and act as a kind of “doctor of the heart” for the tech giant that now refers to itself as an AI-first company.
To advise teams inside Google, Krettek works closely with external organizations.
“I would say part of my creative army is out in the world, which I really love having because I think part of the empathy work is to be deep listeners to nature and be connected with that,” she explained. “I try to spend a lot of time outside, and I like working with people from the outside for that reason, so I kind of think of what I’m doing as a seam between the two.”
By design, Google’s Empathy Lab has a small team meant to embed in other departments to provide them with recommendations for how Google can “bring the humanity into deep learning.”
“I would say everything I’m working on levels to AI. Whether it’s AI sensors through advanced technologies or hardware, there are lots of different elements, but being an AI-first company, it’s like all roads lead to Rome,” she said.
The SXSW panel took place days after Li published a New York Times op-ed about making AI that’s good for people, not just machines.
The audio problem: Learn how new cloud-based API solutions are solving imperfect, frustrating audio in video conferences. Access here