This is the web version of Eye on A.I., Fortune’s weekly newsletter covering artificial intelligence and business. To get it delivered weekly to your in-box, sign up here.
Robots that make pizzas and cappuccinos are having a tough new year.
In recent weeks, two high-profile robotics startups specializing in food preparation, have both significantly slashed costs. Automated coffee shop Café X shut down three San Francisco-based stores, laid off some staff, and is now focusing on two robotic cafes in airports, Axios reported. Meanwhile, Zume Pizza fired over 200 employees and has pivoted from pizza-making robots to creating sustainable packaging for food, Business Insider reported.
Both startups were reportedly attempting to raise additional funding.
The troubles at the two companies are noteworthy because they come at a time when expectations for cutting-edge robotics are high. The rise of machine learning and the declining cost of robotic components like actuators and sensors have led to startups more easily and cheaply building robots that can perform tasks like pick-and-grasp certain objects. Investors and businesses had high hopes that this dual rise of A.I. and cheaper hardware would usher a new robotics age.
So does the problems with these two startups mean that the robot industry is about to go bust?
No, said Remy Glaisner, an analyst at research firm International Data Corporation. The worldwide market for robotics, including sales and related services, will grow 17.8% year-over-year to $112.4 billion in 2020, according to IDC. Manufacturers such as auto makers and aerospace companies will spend more money on robots because advances in A.I. have made it possible for the machines to do more critical jobs, like recognizing which objects they need to pick up and which ones they should avoid, Glaisner explained.
Andra Keay, the managing director of industry group Silicon Valley Robotics, agreed that it’s the more boring uses of robots in industrial or warehouse settings that are on the rise. These kinds of robots can follow warehouse workers, helping them save time zig-zagging around buildings retrieving and restocking inventory.
But some venture capitalists, Keay lamented, like funding sexier robotic startups that they believe will generate news buzz, which they use to lift the valuation of the companies. Indeed, Zume Pizza was valued at $1 billion, according to deal-tracking service Pitchbook.
Keay is critical of the venture capital model, in which startups receive massive amounts of cash in order to grow quickly so that investors can cash out with massive returns after just a few years. This model, she believes, is generally ill-suited for robotics because it takes far longer to develop robots and a business model to go with them.
“Most startups think when they get funded they are getting married, but for most investors it’s just a hookup,” Keay said.
Keay said she urges robot startups to consider funding from universities or industry consortiums instead of venture capitalists. And if they do take money from venture capitalists, they need to “interview investors carefully and make sure that visions are aligned.”
“Sadly the stories so far have focused on robotics not being ready for the real world, when every ‘failed’ robotics company that I know has a story to tell of venture investors screwing them over,” Keay said.
Jonathan Vanian
@JonathanVanian
jonathan.vanian@fortune.com
A.I. IN THE NEWS
Dating A.I. avatars? The rise of deep fakes—or A.I.-generated fake, but realistic images—has led to some companies using the tech “to create imaginary models and ‘increase diversity’ in their ads,” The Washington Post reported. The article said that an unnamed dating-app company plans to use deep fake images “in a chatbot” for unspecified purposes.
Walmart’s new robot helpers. Walmart, through the startup Alert Innovation, has been testing a robotic picking-and-packing system called Alphabot in its Salem, NH store’s warehouse, CNBC reported. “Ultimately, Walmart says Alphabot will cut the time it takes to pick and pack an online grocery order, thereby allowing shoppers to order even more at the 11th hour,” the article said.
A.I.-powered hypochondria. Wired probed cancer-screening technology in the wake of a recent Google research paper that described an A.I. system that performs better than humans at discovering cancer in mammograms. The article describes the limitations of some cancer-screening technology and the possibility that A.I. could exacerbate current problems with cancer-screening tech.
Voluntary regulations. The White House debuted voluntary guidelines for automakers to follow as they develop self-driving vehicles, the Associated Press reported. The article said that some auto-safety experts and the National Transportation Safety Board were disappointed in the voluntary guidelines and “condemned a lack of state and federal regulation for testing autonomous vehicles.”
THE WORST IDEA?
Garry Kasparov, the chess grandmaster who was defeated by IBM’s Deep Blue technology at chess, has harsh words for Hollywood studios using A.I. to help greenlight movies to make, among other feats. “The worst idea since they taught them to play chess!” Kasparov tweeted in response to Warner Bros. using the startup Cinelytic’s predictive analytic technology as part of the studio’s movie-development process. Kaspraov then retweeted a snarky comment about Warner Bros. that said “I’m sure the AI audiences will love the chosen films.”
EYE ON A.I. RESEARCH
A.I. to predict the stock market. Researchers from the Chinese University of Hong Kong and Hebrew University of Jerusalem published a paper that examined deep learning’s performance at predicting stocks. Although the paper describes some of the limitations of A.I., the authors wrote that “machine learning-based investments hold considerable promise for asset management.” A Wall Street Journal article about the paper described how “machine-learning techniques largely failed to live up to their promise of finding previously hidden patterns in the stock market,” but A.I. still has potential despite the enormous amount of hype over the past few years.
FORTUNE ON A.I.
Pharma Needs to Stop Chasing Shiny Objects and Start Embracing Practical Technology—By Paul Hudson
Chip Wars 2020: What AMD, Intel, Nvidia, and Qualcomm Announced at CES—By Aaron Pressman
Facebook Deepfake Video Ban May Set Off ‘Cat and Mouse’ Game—By Alyssa Newcomb
BRAIN FOOD
Feel the chill. The BBC examines the current state of A.I. and the possibility of an upcoming A.I. winter, a period in which expectations and interest in A.I. are low after years of hype. There are some of signs that A.I. is reaching a plateau, as many current A.I. techniques like deep learning are only good at a few, “narrow” tasks, like helping Google’s DeepMind unit create software that defeated human players in the board game Go. But some researchers believe that more A.I. breakthroughs will occur during the next decade, and prefer the term “A.I. autumn” as a description of the state of A.I.
