Google wants to start building artificial intelligence with “people in mind.”
To that end, the search-engine giant is launching the People + AI Research Initiative (PAIR), which will analyze how people interact with A.I. systems; based on that analysis, PAIR will presumably make recommendations about how those systems are designed.
PAIR will explore A.I. in the context of three groups of professionals: engineers and researchers, “domain experts” (i.e., specialists such as doctors or technicians), and everyday users.
“One key to the puzzle is design thinking. Instead of viewing AI purely as a technology, what if we imagine it as a material to design with?” stated Google’s official blog posting on the matter. “Here history might serve as a guide: For instance, advances in computer graphics meant more than better ways of drawing pictures—and that led to completely new kinds of interfaces and applications.”
While the blog’s point seems a bit opaque, it presumably means that advances in A.I. will drive the creation of as-yet-unimagined modes of interaction. But for the moment, at least, many tech companies seem focused on voice and natural language as the best ways for users to tell nascent A.I. platforms what they want.
In its quest to ensure seamless interaction between humans and A.I. platforms, PAIR has released two open-source tools: Facets Overview and Facets Dive, both designed to give engineers clear insight into the data feeding their A.I. algorithms. That can help with debugging and performance monitoring.
Specifically, Overview analyzes feature data from datasets, and visualizes that analysis, offering a faster understanding “of the distribution of values across the features of their dataset(s)” (in the words of the repository’s text). This should illuminate any missing or unexpected feature values that could skew the A.I.’s training.
Facets Dive, meanwhile is an “interactive interface for exploring the relationship between data points across all different features of a dataset.” Like Overview, it’s a visualization designed for easy analysis of an A.I. platform’s inner workings.
But those tools won’t make A.I. platforms friendlier to human interaction, particularly when it comes to everyday users. That will hinge on designers and developers who recognize what people will want out of their software, and build user interfaces accordingly.
For those tech pros developing A.I. applications who also need funding, Google is also rolling out Gradient Ventures, described as a venture fund for “AI-based products.” While the size of the fund hasn’t been publicly announced, and its score remains unclear, it shows that Google is nonetheless interested in the nascent startup community around A.I. When you’re a tech giant, you never know where you’re going to find your next acquisition.