What Google’s New Open-Source Software Means for Artificial-Intelligence Research

Google wants the artificial-intelligence software that drives the company’s Internet searches to become the standard platform for computer-science scholars in their own experiments.

On Monday, Google announced it would turn its machine-learning software, called TensorFlow, into open-source code, so anyone can use it.

“We hope this will let the machine-learning community — everyone from academic researchers, to engineers, to hobbyists — exchange ideas much more quickly, through working code rather than just research papers,” Google announced on its website.

Until now, researchers have had access to similar open-source software: Torch, built by researchers at New York University, as well as Caffe and Theano, are also open to everyone. TensorFlow is meant to combine the best of the three, Jeff Dean, a top engineer at Google, told Wired.

“I think it will be extremely widely adopted by researchers and students in universities and in companies,” Christopher Manning, a computer scientist at Stanford University, told The New York Times. After trying TensorFlow, he described the software as better and faster than the alternatives.

But while TensorFlow is now open-source, Google will continue to manage the project, and some researchers question what that level of control might mean.

“This platform will live or die based on how they handle who controls updates to the code,” Gary Bradski, a computer scientist and president of OpenCV, told the Times. “Can the community have a say, or will Google control the official version by fiat?”

The announcement is “part of a platform ploy,” Oren Etzioni, executive director of the Allen Institute for Artificial Intelligence, told the Times. Google, he said, is trying to attract developers and new hires to its technology.

Return to Top