This is the web version of Eye on A.I., Fortune’s weekly newsletter covering artificial intelligenceand business. To get it delivered weekly to your in-box, sign up here.
Since Bill McDermott became its CEO in November, business software maker ServiceNow has made a couple of notable A.I.-related acquisitions.
In January, the company said it would acquire both Passage AI, which specializes in technology that helps computers understand language, and Loom Systems, a startup that uses machine learning to spot errors in corporate infrastructure for IT staff.
Those two acquisitions came on top of another acquisition by ServiceNow in October of machine learning and natural language processing startup Attivio, which took place just weeks before McDermott officially started on the job in November.
It’s no coincidence that ServiceNow, which sells software to IT professionals but is expanding to other areas, is going on an A.I. acquisition binge, McDermott, the former CEO (and Co-CEO) of business software giant SAP told Fortune. Gobbling up small machine-learning focused startups helps ServiceNow with its plan to embed its products with A.I. as well as obtain A.I. talent in a tight market. Acquisitions are now core to ServiceNow’s A.I. strategy, akin to what Salesforce, Intel, and Apple have been doing in recent years.
“We are continuing to invest in A.I.—we are continuing to invest in deep machine learning and all the predictive analytics that go into the platform,” McDermott said.
Research firm CB Insights said last fall that there have been about 635 A.I.-related acquisitions from 2010 to 2019, underscoring the business world’s increasing interest in A.I. And while tech giants like Facebook, Amazon, and Google are among the biggest acquirers of A.I. startups, CB Insights said “today, smaller AI startups are becoming acquisition targets for traditional insurance, retail, and healthcare incumbents.”
McDermott is especially excited about natural language processing, which has helped ServiceNow create services that can automatically translate languages for its clients. These language-translation features are helpful, he explained, when, for instance, a customer’s IT staff in Germany needs to troubleshoot problems originating from Japan.
By acquiring A.I. companies, ServiceNow can also now tell customers that it’s investing in A.I. at a time when every tech company seems to be doing so.
“No question,” said McDermott.
Still, one problem that often comes up for companies that acquire many smaller businesses is a convoluted product portfolio that may confuse customers. Too many A.I. acquisitions can lead to a puzzling spiderweb of products and services.
To offset any potential problems, McDermott said ServiceNow “immediately” integrated all of the recent technologies from the companies it acquired into its own services. Doing so earlier rather than later is important so that clients aren’t left trying to figure out how each new product works with their own technologies.
Regarding future A.I. acquisitions, McDermott insinuated that more are likely.
As he put it, “You can’t solve 21st century problems with 20th century technology.”
A.I. IN THE NEWS
The DoD chimes in on A.I. ethics. The U.S. Department of Defense introduced its “ethical principles for the use of Artificial Intelligence” after a 15-month period of consulting with “leading AI experts in commercial industry, government, academia and the American public.” The Defense Department’s ethical principles involve A.I. that is responsible, equitable, traceable, reliable, and governable. Lt. Gen. Jack Shanahan, director of the Pentagon’s Joint Artificial Intelligence Center, explained to reporters why the principles lack specific recommendations, government trade publication Breaking Defense reported. “Tech adapts, tech evolves,” the publication quoted Shanahan as saying. “The last thing we wanted to do was put handcuffs on the department to say what we could and could not do. So the principles now have to be translated into implementation guidance.”
A.I.’s coronavirus problem. Some companies that make A.I.-powered health-diagnostic aren’t updating their technology to account for the coronavirus, the Wall Street Journal reported. The article explains that because some of these tools are “built on machine learning,” they “are difficult to tweak for a new virus,” in which data is relatively scant. “Any little-before-seen, highly infectious disease for which data is scarce could pose similar challenges for even the more sophisticated of such technologies,” the article said.
Self-driving cars are not cheap to make. Waymo, the autonomous automobile company that’s a subsidiary of Google parent Alphabet, said it raised $2.25 billion in funding from investors like Silver Lake Partners and the Canada Pension Plan Investment Board. The deal is noteworthy for being the company’s first investment outside of Alphabet.
Get a grip. In recent years, several A.I. startups have emerged that specialize in creating software that helps robotic grippers better pick and place objects then they have in the past. While many of these startups have bragged about their capabilities, tech publication VentureBeat reported that roboticists have yet to agree upon benchmarks that can measure progress in robotic gripping.
EYE ON A.I. TALENT
Intel’s chief of its A.I. platforms group, Naveen Rao, said on Twitter that he is leaving the company, but did not say where is going. Gadi Singer, Intel’s vice president of A.I. products, will take over Rao’s previous responsibilities, tech publication Tom’s Hardware reported. Rao’s departure comes after Intel’s recent $2 billion purchase of A.I. chip maker Habana Labs and its decision to stop producing A.I. chips from its 2016 purchase of Rao’s former startup Nervana Systems.
The Partnership on A.I. (PAI), a nonprofit whose members hail from companies like Apple and Facebook as well as universities and research and legal groups, added three new board members: Rebecca Finlay, the vice president of engagement and public policy for the Canadian Institute For Advanced Research, an organization that does research and charity; Thore Graepel, a research group lead at Google’s DeepMind unit; and Prem Natarajan, the vice president for Amazon’s Alexa A.I. group. The PAI said that Graepel would replace “Mustafa Suleyman’s DeepMind seat on the board,” while Prem would “assume Ralf Herbrich’s previous Amazon seat.”
Sisu, a startup specializing in machine learning and business analytics, hired Berit Hoffmann to be its vice president of product. Hoffmann was previously a product head for Google and its hire and cloud talent service.
EYE ON A.I. RESEARCH
This may be harder than we thought. Researchers from Google published a paper that details some of the challenges of using machine learning technologies to anonymize healthcare patient data, which the researchers said would benefit other researchers. The authors wrote that while some automatic de-identification systems have already surpassed “that of human annotators,” healthcare companies have yet to widely use the technologies because those systems “cannot guarantee performance on all medical text it will ever encounter.” Stat News, a news publication focusing on health, has a good summary of the paper: “Even their best efforts to de-identify health data, or to render it anonymous, would leave some people exposed.”
FORTUNE ON A.I.
In A.I., what would Jesus do?—Jeremy Kahn
Here’s what’s next for a self-driving car startup that’s poised to make history—By Eric C. Evarts
No, A.I. isn’t deciding which movies to green-light—By Dan Reilly
How A.I. helps sniff out new antibiotics—By Sy Mukherjee
College backlash against facial recognition technology grows—By David Morris
Four takeaways from the annual RSA security conference—By Jonathan Vanian
An A.I. dungeon master. A.I.’s next big breakthrough could mean so-called dungeon masters in the classic role-playing game Dungeons & Dragons may be one day replaced by software that generates coherent narratives. Wired examined a recent A.I. research paper detailing a Georgia Tech graduate student’s quest to create a capable A.I. dungeon master. Pulling off the feat would be impressive because a dungeon master must create logical, fantastical narratives that make for fun adventures for human players. The report explained that although current state-of-the-art natural language processing technologies can occasionally produce impressive text far better than previous software, they still create a lot of strange nonsense, indicating that these systems lack a fundamental understanding of language. But further research into A.I. and language, such as the recent paper about dungeon masters, could help humans understand what makes good storytelling, Georgia Tech researcher Lara Martin told Wired.