Venture capital investors can’t get enough of startups focused on artificial intelligence and machine learning, with some $18.4 billion invested in AI startups since 2012. At the same time, it’s an especially tough field for the startups, which face massive competition when they try to scale up for resources such as talent and crucial datasets from tech giants.

Google Inc. today is launching a new program aimed at helping AI startups with those challenges. The Google Developers Launchpad Studio will offer a half-dozen or so AI-driven app startups free access to mentors such as AI expert and Google Director of Research Peter Norvig and Director of Open Source Chris DiBona, services from Google and others, a toolkit that includes datasets and testing tools, and collaboration with customers in various industries.

Startups from three people in a garage to “established post Series B” companies are welcome to apply starting today for the program, which will choose six to eight companies for the first class in October, said Roy Glasberg, the global lead for Google Developers Launchpad. Deadline is Aug. 31.

That program, begun in December 2015 in San Francisco, has hosted four classes of startups from emerging markets in an existing Launchpad Accelerator space in San Francisco (pictured), with hubs in Tel Aviv and New York, and later will expand to Singapore, Bangalore, London and Toronto. The studio will run in parallel with the accelerator program.

In the process of conducting the four accelerator classes, Google found that some startups ran into special challenges. “AI and machine learning startups have very different kinds of needs,” said Glasberg.

Much as Google’s offer of help may sound like the tale of the fox offering the Gingerbread Man a ride across the river before devouring him — Google has acquired 11 AI companies since 2012, the most of any company — Glasberg insisted there are plenty of benefits for startups, investors and prospective customers.

AI and machine learning startups often struggle particularly to figure out the best use case for their technology, Glasberg said, so they need product validation from experienced venture and corporate investors and from users in industry.

In addition, they have trouble hiring talent that often gets scooped up by the big tech companies such as Facebook Inc., Microsoft Corp. and, yes, Google. They also usually don’t have access to datasets that are critical to training machine learning models.

Not least, Glasberg said, startups simply don’t have access to the resources of large companies — even funding, which despite the heady investment in AI isn’t evenly distributed among all the startups out there. And they can benefit from access to AI experts, VCs from the like of Accel Partners and Bessemer Venture Partners, other tech companies such as Intel Corp., and others such as behavioral economist Dan Ariely.

As a result, Google decided it needed to supply a range of services in addition to the mentoring offered in the accelerator program. The idea is to focus on applying AI to specific industries such as agriculture and medicine.

“It’s an awesome learning opportunity for Google and it’s an awesome learning opportunity for startups,” Glasberg said.

That education, of course, also could involve an early look at some of the most promising AI companies. Google announced a new AI and machine learning venture fund, Gradient Ventures, on July 11. Since Gradient said it will provide “capital, resources and dedicated access to experts and bootcamps in AI,” there’s clearly some overlap in the services the two programs are providing.

Google specifically isn’t taking equity or any payment from companies in the studio program, so the two units are independent. But it’s hard to imagine the overriding goal of the studio for Google is to get an early look not only at how AI can be applied to new industries but at the companies and talent it may want to acquire for its own efforts.

The company isn’t alone in this kind of effort, or even first, Glasberg conceded. Element AI Inc., for instance, which raised $102 million in June, is a consultancy that helps organizations develop AI applications. Google just doesn’t want to be left out.

Photo: Robert Hof