M+E Daily

Google Cloud continues to be heavily focused on artificial intelligence (AI) and helping its customers take advantage of all the technology has to offer, according to Google Cloud CEO Diane Greene, who urged companies who haven’t started using AI yet to start doing so because of the major opportunity it presents.

“We’re focused on many, many things at Google Cloud, but we’re particularly focused on helping customers with security and AI” because “security is every company’s biggest threat and AI is every company’s biggest opportunity,” she said Nov. 14 during a keynote presentation during the company’s Global Digital Conference.

Greene pointed out she visits customers all the time to brainstorm how they can use AI to transform their businesses.

“AI has become tremendously useful and Google Cloud [has] really doubled down on making it powerfully useful and super easy to use,” she said. For example, a company can take an application program interface (API) for image recognition and “pop it into your program and suddenly you have this magical” ability to search for and extract images.

But she said: “To really, really exploit AI, it can really require a rethinking of how you do all your processes in a company.”

Despite the advantages offered by AI, people today sometimes “argue that the benefits of this digital age haven’t even fully materialized yet,” she noted. That’s because Google and other “big Internet companies have definitely seen the gains, but it’s just starting to spread out to every industry,” she conceded.

However, she pointed out that if you “look back at the Industrial Age, these things take time” and she predicted “AI is going to take a while too.”

When talking about AI today, we’re usually talking about neural networks, which have been “heavily used” at Google Cloud, she noted, adding: “It’s turned out to be kind of the best solution for a huge variety of problems: things from vision, speech recognition, language understanding, even predictive maintenance.”

She explained that, for a long time, “We’ve just been limited in what we can create. We’ve been constrained by the ability of the computer scientists to devise algorithms to solve the problems, and oftentimes we find that these algorithms just can’t do a good enough job…. You can’t solve things with an algorithm like image recognition or natural language processing. So, rather than using an algorithm, we’ve discovered that we can use a neural network and we can train it to do the things we were trying to hand code these algorithms to do.”

Neural networks have been around for a long time – since the 1980s and 1990s anyway, she pointed out. “But we needed a lot more data [and] a lot more big compute in order to start realizing the breakthroughs we’ve been seeing, and so, in the last 10 to 20 years, we started to have access to a lot of data and huge computer clusters,” she said.

What’s helped significantly is the fact that, today, “there’s over 8 billion connected devices out there and the data is just streaming in” via the Internet, mobile devices and the cloud, she said.

Google is building about one supercomputer a week now to take advantage of AI, she said, adding: “The cloud has really ushered in this new era of AI. It’s really a revolution in how we go after these problems that the algorithms couldn’t solve.”

Summing up her presentation, she said: “Taking advantage of AI requires data. Data is the new oil.” And companies all need to develop a short- and long-term strategy for data that includes consolidating it, getting that data “out of silos, get the most relevant data you can internally and from others [and] getting it processed for the kind of learning you want to do,” she suggested.

She went on to say: “You want to have a long-term vision of where you’re going…. But you can start very small” with AI. “The important thing is to just get going because this AI is, as well all know, for real. It’s revolutionary. It’s going to be incredibly impactful for every business that uses it and we’re here to help you.”

The use of machine learning is “the key thing which means that AI is now viable, where 15 years ago it was always a little bit out there [in] the future,” Google Cloud VP of AI Andrew Moore said in the second keynote presentation.

If an organization has the systems in place allowing them to gather data, they “can build out models which allow your organizations to predict what’s next or predict what’s wrong,” he said, adding: “Where I see us going next is there is actually more to AI than that and I’m really critically interested in what you do with the predictions from the machine learning systems” that are set up.

He predicted we’re “going to see a huge growth in what I call the top of the AI stack in the decision-making systems which are actually tied to what the machine learning system is doing.” That’s the “next big area of growth,” he guessed.

He’s glad, meanwhile, that general AI continues to be researched, he told viewers, explaining: “It may come to something. At the moment, there is so much that we can do to,” as just two examples, “increase safety of people in the world and increase the productivity,” he said.

But he conceded: “AI is currently very, very stupid. It is really good at doing certain things which our brains can’t handle, but it is not something which we could trust to do…things like analogies or creative thinking or jumping outside the box.”

Asked about widespread concerns of job disruption potentially caused by AI, he said: “We all have a duty to be thinking about this.” That’s why, “when you’re introducing a new technology and you get a choice between using it to empower your current workers and make them more productive and enjoy their jobs more, versus just replacing them with a box, we very much push towards the former because that’s usually where you get the successes,” he said.

In a presentation before the keynotes, Stephanie Wong, a Google scalable developer advocate, noted that companies today are “looking for tools that can help jumpstart their AI journeys — whether they have in-house AI expertise or have developers focus on application logic and need AI tools that are more out of the box.” Google Cloud offers “a spectrum of AI tools that supports a variety of these landscapes” and “one goal of the conference was to provide people with an “action plan to incorporate data and AI solutions into your business right now,” she said.

She provided companies with a tip: “If you are trying to find the best place to get started with AI in your business, but haven’t started yet, you might want to consider [that] every good AI strategy starts with a good data strategy.”

Mark Mirchandani, another Google scalable developer advocate, told viewers: “AI is impossible without large volumes of relative business data, regardless of where you are on your journey. Organizations vary, from having a large amount of untapped data and they’re looking to clean it up, transform their data into a state that supports deeper analysis, to those with more structured, analysis-ready data that are doing traditional statistical, historical analyses but are looking to move more into predictive analysis and AI.”

Google Cloud’s platform of services and products can help a company, “regardless of where you are on your data journey,” he said.