SAN FRANCISCO—Democracies, we have found out, can be challenging. They are also full of potential, which may be why the term is being invoked so often by major cloud vendors about the topic of artificial intelligence.
During the Google Cloud Next ’17 conference this week Fei-Fei Li, newly hired chief scientist of artificial intelligence and machine learning at Google Cloud, stood beneath a background that read “Democratizing AI” as she talked about how far humans have come in machine learning and AI—and how far they have to go.
“The next step for AI must be democratization,” she said. This means “lowering the barriers of entry, and making it available to the largest possible community of developers, users and enterprises.”
By my reckoning, “Democratizing AI” was first coined by Salesforce.com CEO Marc Benioff at last year’s Dreamforce, when the company rolled out its Einstein AI apps. However, Benioff has used the term in the past when discussing software-as-a-service. He talked about AI democratization on March 6 when Salesforce announced a partnership with IBM around its own “cognitive computing” AI brand, Watson.
Microsoft CEO Satya Nadella also has been touting the democratization of AI. He wants to put machine learning, image and video recognition and natural language processing “in the hands of every developer, every organization, every public sector organization around the world.”
This is a large task and responsibility, but Google executives feel their AI tools and services, coupled with the power and reach of the Google network, have the potential to change the world.
“Big data is so powerful that nation states will fight over how much data matters,” said Eric Schmidt, chairman of Google parent Alphabet, who is a late convert to the AI dogma. “I’ll bet the rest of my professional career that the future of your business is big data and machine learning applied to the business opportunities, customer challenges and things before you.”
For its part, Amazon Web Services also has its own AI services, rolling out several at its Re:Invent conference last year. But this week it was Google’s turn to evangelize.
Li announced the general availability of Google’s Cloud Machine Learning Engine, which is designed for data scientists to build custom models using TensorFlow, the open source set of machine learning libraries that Google open sourced in 2015.
New to Google’s lineup of machine learning APIs is the Cloud Video Intelligence API, which was released in private beta. The API can detect shapes and objects in videos so they can be tagged and searched.
For instance, a developer could create an application that could search for a corporate logo or product photo in a set of videos, such as Google and YouTube’s set of 8 million videos.