The Complete Magazine on Open Source

Google initiates PAIR to focus on ‘human side’ of AI

Google PAIR AI initiative

While artificial intelligence (AI) was already revolving around humans, Google has now launched its all new initiative called PAIR to study the ways people interact with AI. The latest move brings open source tools to help researchers develop new AI experiences.

“The goal of PAIR is to focus on the ‘human side’ of AI: the relationship between users and technology, the new applications it enables, and how to make it broadly inclusive,” Google’s Senior Staff Research Scientists Martin Wattenberg and Fernanda Viégas wrote in a blog post.

Google is inherently supporting machine learning and improving its position in the AI world by bringing more accurate speech recognition, better image search and improved translations for a long time. But since human interactions are vital to succeed with an AI system, the search giant has now been working on PAIR, which is a “People + AI Research” approach.

Research through PAIR will be broadly divided into three areas, namely engineers and researchers, domain experts and everyday users. While engineering teams would leverage the latest initiative to ease their AI developments through tools and guidelines, augment professionals, doctors, technicians, designers, farmers and musicians will receive appropriate solutions through some operational PAIR research.

Google has dedicated a team of 12 engineers to kickstart developments under PAIR. Moreover, it brought open source visualisation tools to begin developer engagements for the newest initiative.

Open source developments

Known as Facets Overview and Facets Dive, the two open source developments are specifically designed for AI engineers with an aim to “address the very beginning” of machine learning. “The Facets applications give engineers a clear view of the data they use to train AI systems,” describe PAIR researchers.

Facet Overview takes input feature data from datasets to analyse them feature-by-feature and visualise the analysis. Facet Dive, on the other hand, helps to interactively explore a large number of data points at once.

The first open source Facet series is just beginning. Google has plans to bring more open source tools in the future. Also, new grants and residencies will be set up to support AI research. Harvard researcher Prof. Brendan Meade and MIT expert Prof. Hal Abelson are additionally working with the PAIR team to intensify the advanced developments.