Google this week announced a new initiative aimed at making artificial intelligence “more useful to all of us”—by designing more user-friendly systems.
The objective, according to Google Brain Team research scientists Martin Wattenberg and Fernanda Viégas, is to focus on the “human side” of the platform.
PAIR’s research is divided into three areas: engineers and researchers (what educational materials, practical tools are needed?), domain experts (how can AI support doctors, technicians, designers, farmers, musicians?), and everyday users (can we democratize the technology behind AI?).
“We’re open sourcing new tools, creating educational materials (such as guidelines for designing AI interfaces), and publishing research to answer these questions and spread the power of AI to as many people as possible,” Viégas and Wattenberg wrote in a blog post.
Google’s plan to “democratize AI”—make it accessible to end-users and developers—really kicked off at May’s I/O conference, where CEO Sundar Pichai stressed the company’s shift to an AI-first approach.
Machine learning algorithms already influence the ranking of search results and Gmail’s “smart reply” feature, among other mobile and desktop functions. And continues to extend its reach with Google Lens (vision-based computing capabilities in Assistant and Photos) and TensorFlow (open-source machine learning framework for developers).
Those efforts expanded even further on Monday, with the open sourcing of two visualization applications—Facets Overview and Facets Dive—that give engineers “a clear view” of their AI training data, “a key ingredient in modern AI systems,” according to Google.
The tech titan is not, of course, the first to explore human/AI interaction; Wattenberg and Viégas acknowledged those who have come before, adding that “their work inspires us” and is “an essential part of our mission.”
As are the pair of visiting professors—Brendan Meade of Harvard University and Hal Abelson of MIT—who are focused on education and science in the age of AI.
“We don’t have all the answers—that’s what makes this interesting research,” the blog said. “But we have some ideas about where to look.”
Let us know what you like about Geek by taking our survey.