Google wants to PAIR-up AI with human interaction by being open-source by design

Google's PAIR project aims to develop open-source tools aimed at accelerating the use-case, and spreading the uptake of AI / Pixabay

Google has today (July 10) unveiled a research program geared towards improving how humans interact with artificial intelligence (AI). The latest initiative includes open-source tools that will be made available to third parties, including software engineers, as well as professionals from disciplines such as medicine, design and music.

Dubbed PAIR (People + AI Research Initiative), the scheme will aim to spread the use of AI research beyond the areas where it has been pioneered to date, such as speech and image search, by focusing on "the human side of the relationship between users and technology" with a view to developing new applications for such software.

PAIR has opened with the launch of two open-source tools named Facets Overview and Facets Dive — visualization tools aimed at giving software engineers a clear view of the data they use to train AI systems.

According to Google, the project will revolve around aiding three types of users:

  • Engineers and researchers: Providing such practitioners with tools to better understand and build machine learning systems.
  • Domain experts: Exploring how AI can help professionals from all disciplines, including medical, design, farming, etc, use AI.
  • Everyday users: Ensuring the benefits of machine learning are inclusive “so everyone can benefit from breakthroughs in AI”.

A blog post announcing the launch of the project attributed to Martin Wattenberg and Fernanda Viégas, senior staff research scientists at Google Brain, claims that Google wants to extend the current use case of AI by “building systems with people in mind at the start of the process” – a technique it has termed as “human-centered machine learning”.

The post goes on to explain how PAIR is geared towards repositioning the field of AI research from a tech-focused discipline, by introducing elements of “design thinking.”

The pair go on to state: “Advances in computer graphics meant more than better ways of drawing pictures — and that led to completely new kinds of interfaces and applications.”

Commenting on the launch of the open-source tools Facets Overview and Facets Dive, the pair went on to note that “training data is a key ingredient in modern AI”, as machine learning engineering differed from more traditional software engineering due to the stronger need to debug”.

AI has been one of the buzz phrases of the digital marketing sector in recent years, with the newly-minted chief executive of Publicis Groupe Arthur Sadoun stealing the headlines at Cannes Lions 2017 by declaring his intention to eschew entering awards ceremonies the next 12 months — as well as the flagship festival in 2018 — and instead focusing on the development of its AI-powered platform Marcel.

SaveSave

SL

Sean Larkin

All by Sean