Eye on A.I.— Celebrating the Godfathers of Deep Learning

Artificial intelligence’s growing importance, in both business and research, is largely because of Yoshua Bengio, Geoffrey Hinton, and Yann LeCun—otherwise known as the “godfathers of deep learning.”

Last week, the trio won the annual Turing Award, the technology world’s equivalent to the Nobel Prize, presented by the Association for Computing Machinery. Their research has led to huge breakthroughs in computers translating languages and identifying objects and people in photographs.

At the heart of the men’s research are neural networks, the software created decades ago to mimic how the human brain learns. The technology plays a huge role in artificial intelligence and its close cousin, deep learning.

The analogy of neural networks being like the human brain is something many A.I. researchers and neuroscientists loathe because it oversimplifies an extremely complicated process. “That’s okay,” Bengio, a University of Montreal computer science professor and co-founder of enterprise startup Element AI, said in an interview with Fortune a day before his Turing Award was publicly announced.

For years, the A.I. community ignored neural networks in favor of other techniques that are more closely related to conventional computer programming. But more advanced computer chips and access to Internet data supercharged the power of neural networks, which could now be fed enormous amounts of information so they could identify patterns and, thereby, learn.

“Because as we train these systems with more data, they just get better,” Bengio said.

He recalled how difficult it was to get funding when he initially studied neural networks. Ultimately, the Canadian government kicked in some money. Today, in contrast, artificial intelligence attracts big money from governments and venture capitalists, with most of the investment in the U.S. coming from businesses. Bengio only hopes that other technologies, especially those currently lacking buzz, also have easy access to cash.

“We need to be careful to nurture that spirit and not try to say, ‘That’s the end of it,'” he said.

LeCun, now Facebook’s chief A.I. scientist, also talked with Fortune about his early research into neural networks. During those experiments, he said he focused more on creating computers that “think” rather than how the technology would be applied in real life. But now LeCun believes big breakthroughs are imminent, mostly from businesses that specialize in areas including medical imaging, self-driving cars, and even garden maintenance (i.e. lawnmowers that know to trim only weeds instead of roses). What’s certain is that companies will use the technology in ways he never envisioned.

Says LeCun, “It’s not like we have a monopoly on good ideas.”

Jonathan Vanian

Sign up for Eye on A.I.


The equivalent of 60 million Big Macs. McDonald’s plans to pay over $300 million for Israeli machine-learning startup Dynamic Yield, reported the Wall Street Journal. The acquisition's data crunching technology will be used to improve McDonald's digital displays for drive-through windows so that they highlight different food such as ice cream when the weather is warm.

Is this A.I. okay to use? Google announced last week the creation of an advisory council to provide perspective about using A.I. technologies like facial-recognition software ethically. The council, which will meet four times annually, is made up of outside experts from fields like public policy, foreign policy, privacy, and computer science. Since the announcement, Bloomberg News reported that the council has faced some setbacks, including one council member dropping out for unspecified reasons and a group of over 300 Google employees publicly petitioning for the removal of another member who has "fought against equal-rights laws for gay and transgender people."

Cramming A.I. tools into containers. Amazon has debuted a new service for building A.I.-infused apps, reported tech news publication GeekWire. The service relies on containers, a trendy technology that developers use to create complex apps like Uber's, and incorporates some of Google's TensorFlow A.I. tools.

The Vatican has some questions about A.I. The Vatican is concerned about A.I.’s potential impact on society and the possibility that cutting-edge technology will lead to “spiritual malaise,” reported religious news publication Crux. “People could then see themselves no longer as meaningful agents, but rather as passive objects of manipulation by machines or elites with power,” Father Ezra Sullivan told the publication.


Data scientists used to be one of the hottest jobs in business, but now they're rebranding themselves “machine learning engineers,” according to tech publication Datanami. One reason for the job title change: Machine learning engineers typically earn higher salaries.


Enterprise software company Workday has hired Sheri Rhodes as its new CIO, replacing Diana McKenzie. Rhodes was previously the CTO of Western Union.

President Donald Trump has nominated Michael Kratsios as the nation’s CTO. Kratsios was previously the deputy U.S. CTO and principal and chief of staff for Thiel Capital, founded by investor Peter Thiel.

Dr. Ivan Venzor has left Mexican beverage giant FEMSA to become the deputy director of data science and machine learning at Banregio Bank. Venzor is also an A.I. advisor for Mexico’s Industrial A.I. Research Centre in Nuevo León.


A.I. to prevent big blackouts. Researchers affiliated with the Institute of Electrical and Electronics Engineers trade group published a paper about the use of reinforcement learning—in which computers learn through trials—to control power systems during emergencies. If a storm damages a major energy grid, for example, A.I. may be able to troubleshoot any problems faster than human operators.

A.I. to detect melanoma. Researchers from both New York University and its School of Medicine published a paper about their work building a deep-learning system that can detect and classify skin lesions better than other similar systems. The researchers were able to achieve better results by removing certain images like human hair from photos of skin lesions that could confuse deep learning systems.


Finland Is Using Inmates to Help a Startup Train Its Artificial Intelligence Algorithms – By Alyssa Newcomb

Huawei's Perception Problem Deepens as U.K. Spies Identify Security Risks – By David Meyer

UPS Begins Using Drones to Transport Medical Samples at North Carolina Hospitals – By Kevin Kelleher

Trump Tweets About His Off-the-Book Meeting With Google's CEO – By Alyssa Newcomb


The business of your face. Fortune’s Jeff John Roberts explores the business of selling facial-recognition software and discovers some unsettling ways companies are using online photos to train their software. One startup, Ever AI, originally focused on creating an app for people to store their photos, but eventually pivoted to selling companies, law enforcement, and other organizations technology for scanning and analyzing faces. As a Georgetown University professor and face-recognition expert explains, “Unlike fingerprints, where there have long been rules on how and when they’re collected, there are no rules for face technology.”

Subscribe to Well Adjusted, our newsletter full of simple strategies to work smarter and live better, from the Fortune Well team. Sign up today.

Read More

Artificial IntelligenceCryptocurrencyMetaverseCybersecurityTech Forward