Making the artificial feel more natural is one of the goals of computer scientists these days.
Last week in New York, innovators from IBM, Foursquare, Microsoft, and other companies discussed what may lie ahead in the commercial development of artificial intelligence. From baby steps in machine learning to longer strides in interacting with humans, AI has gotten better at interpreting the world—though the technology is still decades away from some of the stuff of science fiction.
NYU Future Labs and ff Venture Capital co-hosted last Tuesday’s forum, which led off with a keynote address from Minerva Tantoco (pictured above), senior adviser with Future\Perfect Ventures. She started her first AI company while still in college, and her early efforts in the field included getting a computer to parse a sentence in English into its grammatical parts. “It was very early days,” she said, “some of the building blocks of the translation software we think is normal today.”
In fact, there are apps available now, Tantoco said, that can help people translate and conduct conversations across two languages. “That is the miracle of AI,” she said.
Until recently, Tantoco served as CTO in Mayor Bill de Blasio’s administration—a role created two years ago. She stepped down in late July to join Future\Perfect Ventures.
The development of AI and related innovations goes beyond the technology, she said, and is tied to how people would use it. “You think about data being objective, but it’s our interpretation of it that is subjective,” Tantoco said.
John Frankel, a partner with ff Venture Capital, moderated the panel consisting of Max Sklar, machine-learning engineer with Foursquare; Maya Weinstein, senior design lead for IBM Watson; Drew Austin, co-founder and CEO of Wade & Wendy; and Ben Tamblyn, manager of corporate communications with Microsoft.
Thanks to the wild imaginings of moviemakers, AI sometimes gets cast as a villainous force hell-bent on conquering the world. The reality is far different, with artificial intelligence already at work in everyday life, according to the panel. “AI is not this big singularity, Skynet kind of creation that’s going to come about one day,” said Frankel. “It’s happening all around us in thousands of different ways.” That includes maps, traffic light controls, and intelligent assistants, he said.
Making sure AIs stay on some sort of moral path, though, rather than turn on their creators, is on the minds of developers including the folks at IBM. “Watson is not meant to replace humans,” Weinstein said. “It is meant to be an assistant to a human. It helps you through a specific process you are struggling with.”
IBM’s Watson platform, which uses machine learning to form insights from big data, has found a role as a medical assistant, she said. “It went to medical school and learned everything that medical students learn, went through all of the tests and training, graduated, and now it is being used as an oncology expert and for clinical trial matching,” Weinstein said. It helps doctors with their diagnoses, but is not there to take over. “Watson has a terrible bedside manner and cannot deal with certain situations,” she said.
Watson is also familiar with crunching data from the sports world, and was recently put through its paces again at this year’s US Open tennis championship.
Austin said the AI work being done in New York is an opportunity to further elevate the city’s presence among innovation hubs around the country. “What San Francisco had for engineers, New York could have for AI and data scientists,” he said. His startup, Wade & Wendy, got its name from a pair of AIs in development to help in the recruiting and hiring sector. The company is focused on “conversationally intelligent” AI, Austin said, which gets to know job seekers, their skills, interests, and goals—and then presents content and career opportunities relevant to them.
At its root, the intent of artificial intelligence is to process information and present new ideas and perspectives on what it all may mean. Sklar said Foursquare analyzes the tons of data collected from people checking in through its app. “A lot of our focus initially was to take all the machine learning algorithms we learned in school and apply them,” he said. What Foursquare discovered, Sklar said, was that
