Artificial IntelligenceCryptocurrencyMetaverseCybersecurityTech Forward

Artificial Intelligence Hype Meets Reality: Eye on A.I.

July 2, 2019, 3:01 PM UTC

Despite all the hype about artificial intelligence, most executives expect that it will take years before the cutting-edge technology gives their businesses a financial lift.

Their long-term view was laid out by consulting firm KPMG in a recent survey of 400 executives, all of whom had artificial intelligence projects in progress within their companies. The executives generally said that it would take some time before artificial intelligence pays dividends, signaling a growing realization that the technology won’t have the quick impact that many had initially hoped for.

Just over half of the executives surveyed, 51%, said it will take three-to-five years before their A.I. projects create a “significant return on investment.” That’s in sharp contrast to last year’s survey, in which only 28% said it would take that long—highlighting how much executives have reconsidered their initial rosy expectations.

Meanwhile, 47% of respondents said they expect significant results in three years or less. That marks a major decline from last year, when 62% expected short-term results.

Brad Fisher, a KPMG partner and U.S. leader for data and analytics, says executives increasingly realize that machine-learning projects take time to pay off. Fisher said that, anecdotally, CEOs “are upbeat and optimistic, but recognizing that it’s harder then you think,” implying executives aren’t frustrated by the delays, at least not yet.

Fisher believes that executives are still excited about cutting-edge data crunching even through many of their older data-analytics projects didn’t go as well as many of them had expected.

A decade ago, many executives were excited about a different buzzy technology, big data, that involved collecting and crunching massive amounts of information without quite the complex analysis that artificial intelligence promises. Despite the hype at the time, many big data projects failed to deliver significant returns for companies, Fisher said.

Some companies, he said, spent as much as $60 million to install big-data infrastructure before deciding how it would be specifically applied, like predicting sales of a particular product. As a result, many projects stalled while the data that companies collected “just sat there.”

In a sign that business leaders have learned their lessons, Fischer said that many of them are deploying machine learning only after first determining how it will be used. For example, executives hope it can identify the customers that salespeople should target or to forecast a company’s annual revenue, he explained.

Still, like with nearly every new technology, reality is finally and inevitably setting in with artificial intelligence. The initial euphoria is ebbing. Now it’s all about hard work and patience.

Jonathan Vanian

Sign up for Eye on A.I.


Take this down. The creator of the controversial DeepNude app has taken it offline after fierce backlash, tech publication Motherboard reported. The app, which uses A.I. to automatically alter images of women so that their clothing appears to be removed, was heavily criticized by a number of experts, including deep-learning pioneer Andrew Ng.

Not reliable enough. Axon, the law enforcement technology and weapons maker formerly known as Taser, said that it would “not be commercializing face matching products on our body cameras at this time,” citing “serious ethical concerns” and “technological limitations.” The company cited a report by its A.I. ethics and policy board as a determining factor and said that its “AI team will continue to evaluate the state of face recognition technologies and will keep the board informed about our research.”

Goodbye manufacturing jobs. CNN reported on an Oxford Economics study that estimates that 20 million manufacturing jobs worldwide could be eliminated because of automation technologies and robotics. Even if robotics and automation end up creating new jobs, there’s a possibility that the technologies “could contribute to income inequality.”

Hello machine-learning jobs. A survey from employment-search service Indeed showed that A.I.-related job postings grew 29% year-over-year at the end of May, reported Bloomberg News. The report said that “employers in fields as diverse as media, finance and medicine are searching for machine learning engineers to help transform and enhance their product offerings.”


A former employee of Dynamic Yield, a data-crunching startup acquired by McDonalds for $300 million, wrote a piece on Medium about startups claiming they use “A.I.” in their products. Mike Mallazo explains that while Dynamic Yield’s technology will be a useful tool for McDonald’s to better understand its customers, it’s not actually using A.I. Mallazo writes: “Yet in my two years at the startup, reporters, analysts, and sometimes even customers seemed determined to call us an A.I. company. For a while, we resisted the A.I. label, understanding that our platform wasn’t going to make Watson sweat anytime soon. But eventually, we gave up and just decided to kind of go along with the hype. The market wanted us to be an A.I. company so we chuckled and decided to call ourselves one.


Self-driving automobile startup Waymo hired a team of 13 robotics experts from the recently shuttered toy robot maker Anki. The new team, which includes Anki’s former CEO Boris Sofman, will help Waymo “lead its nascent trucking initiative,” Axios reported.

Cubic Corporation picked Jim Colson to become the defense and transportation company’s vice president and chief technology officer. Colson was previously an IBM fellow and vice president, and CTO of the Watson customer engagement unit.

Healthcare startup CureMatch hired Viktor Novy to be the company’s CTO. Novy was previously the IT director of healthcare company Human Longevity.


A.I. to detect fungal infections. Researchers from Jagiellonian University in Poland published a paper about using deep learning to analyze microscopic images of different type of fungus species. The researchers claim that their A.I. techniques could shorten the amount of time it would take healthcare professionals to select the correct antifungal drugs to treat patients.

Preventing A.I. spoofs. Researchers from Australia’s Commonwealth Scientific and Industrial Research Organisation published a paper detailing techniques to safeguard A.I. systems from being spoofed by so-called adversarial examples, or images. Adversarial images are deep-learning altered images and photos that contain certain elements undetectable to the human eye that cause computers to stumble when trying to identify them.


Apple’s Latest Acquisition Shows Self-Driving Cars Are in the Doldrums of Disappointment – By David Z. Morris

AMD Denies Improperly Giving Sensitive Chip Technology to China – By Aaron Pressman

Mark Zuckerberg: Facebook is Working on a Policy for Policing Deepfake Videos – By Danielle Abril


Don’t forget classical music. The New York Times reports on how classical music often gets lost in the mix by the music-recommendation algorithms of popular music streaming services Spotify, Apple, and Amazon. These services tend to recommend pop music to listeners, partly because the algorithms were built to accommodate so-called metadata like song titles, album names, and track listings. The algorithms fail at recommending or accurately retrieving classical music because the genre often contains “movements” and “composers” and other ways to characterize data that are different than the pop genre. “If you have Herbert von Karajan conducting a Verdi opera with Maria Callas, who is the artist?” Idagio executive Till Janczukowicz tells the Times.