AI Hiring Frenzy Settles Down

Instead, tech companies are shifting the focus of AI hiring toward engineering and product groups building everything from improved ad products, virtual assistants, image-recognition and recommendation engines, and have reassigned some of their existing AI researchers to those areas.

“AI is in the process of maturing from academic and basic research to niche applications, to wide deployment,” said Yann LeCun, vice president and chief AI scientist at Facebook. “As the field matures and as the tools become better, companies are massively increasing their investment in the engineering, development, tooling and infrastructure related to AI.”

Mr. LeCun said Facebook, for its part, isn’t cutting back on investing in fundamental research. “In a mature field, the size of basic research is always much smaller than engineering and product development.”

While big tech companies still have job openings for AI researchers, the days of teams doubling and tripling in size appear to be over, if only because of how large the groups have become. Microsoft’s AI and Research Group, for instance, jumped to 8,000 people from 5,000 between 2016 and 2017. As The Information reported in mid-2017, companies began hiring academics in part so they could have an easier time hiring their students when they graduated. 

AI encompasses a vast array of technologies, including software that can recognize people and objects in images and video, and understand the meaning of human speech. Most of the growth has occurred in a branch of AI called “deep learning,” in which computers analyze large amounts of data to identify meaningful patterns. Many of the pure AI researchers hired in recent years worked with mountains of data generated by their companies conducting experiments and developing new algorithms. Some in the field say the developments in deep learning that allowed its use in specific products is partly why the pace of hiring has changed.

While he wasn’t aware of staffing trends at Microsoft and Google, Ed Lazowska, a computer science professor at the University of Washington, said it’s understandable that companies would shift AI talent to products given the limited number of experts in the field and the abundance of “low-hanging fruit”— near-term opportunities to boost profits through new technologies.

“Given the current gold rush, it’s only natural that companies are going to deploy their available talent to harvest this low-hanging fruit rather than to push the frontiers of AI in ways that may not pay off for a number of years,” he said.

Google and Microsoft

At Google, regarded as the tech industry’s largest pool of AI research expertise, hiring for AI research roles “has certainly slowed” in terms of percentage growth over the past three years, according to a person who works at Google. Part of the reason is that several of Google’s AI research teams—including Google Brain, widely considered one of the crown jewels of Alphabet’s AI research work—have seen a significant expansion over that period, causing their relative hiring growth rates to fall, said the person.

Alphabet has also restructured its AI units. In April 2018, Google appointed Senior Fellow Jeff Dean, who was previously the chief of Google Brain, to lead all of Google’s AI efforts after the company’s former AI chief John Giannandrea left to join Apple. Last November, Alphabet moved the health AI unit of DeepMind, its U.K.-based AI subsidiary, into Google to be part of a new health AI unit that includes Google AI researchers.  

Google asked some of its existing AI researchers to switch to positions with product teams, according to two AI industry executives with contacts at Google.

Google has also been assigning Google Brain people to work with AI Hub, a Google Cloud unit announced in November that is focused on providing prebuilt AI software designed for companies to plug into their operations, according to a person who does business with Google.

“We’re bringing together many people from across the company to work on our AI efforts, and we’re also training more and more people to do this sort of work so, for example, more and more engineers and researchers have gone through our internal machine learning training,” said a person close to Google.

At the same time, Alphabet has in recent months started emphasizing the term Google AI as an umbrella term for its AI research and product-related work. One result is that Google is using the term “Google Brain” less than in the past—a reflection of the fact that Alphabet has many different AI teams, said the person close to Google. A person who works at Google said the idea is to avoid using “territorial” brands for the company’s AI work.

Meanwhile, Microsoft Research, which leads much of the software giant’s AI research, is scaling back on AI research hiring, according to two AI researchers outside of Microsoft who have been briefed on the situation by peers at the company.

A person close to Microsoft said as its AI efforts have matured, the company’s needs for certain types of personnel and skill sets have changed. This has resulted in “shifting” headcounts for certain AI research projects, as well as a need for AI research teams “to work within a finite budget,” the person said. Microsoft’s overall commitment to AI hasn’t changed, the person said.

Microsoft Research, founded in 1991 as a place for computer scientists and researchers to develop new technology, began shifting some of its AI researchers closer to product teams several years ago. In 2016, Microsoft moved Microsoft Research into its newly formed “AI and Research Group”—which included Cortana, Bing and Microsoft’s robotics work—and appointed Microsoft Research EVP Harry Shum to lead it.

—Nick Wingfield contributed to this article