‘Smile with your eyes’: How to beat South Korea’s hiring bots

A three-hour class in AI hiring techniques can cost more than $80. (Reuters)
Short Url
Updated 17 January 2020

‘Smile with your eyes’: How to beat South Korea’s hiring bots

  • As many as eight out of every 10 South Korean students are estimated to have used cram schools

SEOUL: In cram school-obsessed South Korea, students fork out for classes in everything from K-pop auditions to real estate deals. Now, top Korean firms are rolling out artificial intelligence in hiring — and jobseekers want to learn how to beat the bots.

From his basement office in downtown Gangnam, careers consultant Park Seong-jung is among those in a growing business of offering lessons in handling recruitment screening by computers, not people. Video interviews using facial recognition technology to analyze character are key, according to Park.

“Don’t force a smile with your lips,” he told students looking for work in a recent session, one of many he said he has conducted for hundreds of people. “Smile with your eyes.”

Classes in dealing with AI in hiring, now being used by major South Korean conglomerates like SK Innovation and Hyundai Engineering & Construction, are still a tiny niche in the country’s multi-billion dollar cram school industry. But classes are growing fast, operators like Park’s People & People consultancy claim, offering a three-hour package for up to 100,000 won ($86.26).

There’s good reason to see potential. As many as eight out of every 10 South Korean students are estimated to have used cram schools, and rampant youth unemployment in the country — nearly one in four young people are not in the workforce, according to Statistics Korea — offers a motive not present in other countries where cram schools are popular, like Japan.

Businesses around the world are experimenting with increasingly advanced AI techniques for whittling down applicant lists.

But Lee Soo-young, a director of Korea Advanced Institute of Science and Technology (KAIST) Institute for Artificial Intelligence, told Reuters by telephone the new technology is being more widely embraced in South Korea, where large employers wield much influence in a tightening job market.

According to Korea Economic Research Institute (KERI), nearly a quarter of the top 131 corporations in the country use or plan to use AI in hiring.

One AI video system reviewed by Reuters asks candidates to introduce themselves, during which it spots and counts facial expressions including “fear” and “joy” and analyzes word choices. It then asks questions that can be tough: “You are on a business trip with your boss and you spot him using the company (credit) card to buy himself a gift. What will you say?”

AI hiring also uses “gamification” to gauge a candidate’s personality and adaptability by putting them through a sequence of tests.

“Through gamification, employers can check 37 different capabilities of an applicant and how well the person fits into a position,” said Chris Jung, a chief manager of software firm Midas IT in Pangyo, a tech hub dubbed South Korea’s Silicon Valley.


Man vs. machine in bid to beat virus

Updated 20 February 2020

Man vs. machine in bid to beat virus

  • Human and artificial intelligence are racing ahead to detect and control outbreaks of infectious disease

BOSTON: Did an artificial-intelligence system beat human doctors in warning the world of a severe coronavirus outbreak in China?

In a narrow sense, yes. But what the humans lacked in sheer speed, they more than made up in finesse.

Early warnings of disease outbreaks can help people and governments to save lives. In the final days of 2019, an AI system in Boston sent out the first global alert about a new viral outbreak in China. But it took human intelligence to recognize the significance of the outbreak and then awaken response from the public health community.

What’s more, the mere mortals produced a similar alert only a half-hour behind the AI systems.

For now, AI-powered disease-alert systems can still resemble car alarms — easily triggered and sometimes ignored. A network of medical experts and sleuths must still do the hard work of sifting through rumors to piece together the fuller picture. It is difficult to say what future AI systems, powered by ever larger datasets on outbreaks, may be able to accomplish.

The first public alert outside China about the novel coronavirus came on Dec. 30 from the automated HealthMap system at Boston Children’s Hospital. At 11:12 p.m. local time, HealthMap sent an alert about unidentified pneumonia cases in the Chinese city of Wuhan. The system, which scans online news and social media reports, ranked the alert’s seriousness as only 3 out of 5. It took days for HealthMap researchers to recognize its importance.

Four hours before the HealthMap notice, New York epidemiologist Marjorie Pollack had already started working on her own public alert, spurred by a growing sense of dread after reading a personal email she received that evening.

“This is being passed around the internet here,” wrote her contact, who linked to a post on the Chinese social media forum Pincong. The post discussed a Wuhan health agency notice and read in part: “Unexplained pneumonia???”

Pollack, deputy editor of the volunteer-led Program for Monitoring Emerging Diseases, known as ProMed, quickly mobilized a team to look into it. ProMed’s more detailed report went out about 30 minutes after the terse HealthMap alert.

Early warning systems that scansocial media, online news articles and government reports for signs of infectious disease outbreaks help inform global agencies such as the World Health Organization — giving international experts a head start when local bureaucratic hurdles and language barriers might otherwise get in the way.

Some systems, including ProMed, rely on human expertise. Others are partly or completely automated.

“These tools can help hold feet to the fire for government agencies,” said John Brownstein, who runs the HealthMap system as chief innovation officer at Boston Children’s Hospital. “It forces people to be more open.”

The last 48 hours of 2019 were a critical time for understanding the new virus and its significance. Earlier on Dec. 30, Wuhan Central Hospital doctor Li Wenliang warned his former classmates about the virus in a social media group — a move that led local authorities to summon him for questioning several hours later.

Li, who died Feb. 7 after contracting the virus, told The New York Times that it would have been better if officials had disclosed information about the epidemic earlier. “There should be more openness and transparency,” he said.

ProMed reports are often incorporated into other outbreak warning systems. including those run by the World Health Organization, the Canadian government and the Toronto startup BlueDot. WHO also pools data from HealthMap and other sources.

Computer systems that scan online reports for information about disease outbreaks rely on natural language processing, the same branch of artificial intelligence that helps answer questions posed to a search engine or digital voice assistant.

But the algorithms can only be as effective as the data they are scouring, said Nita Madhav, CEO of San Francisco-based disease monitoring firm Metabiota, which first
notified its clients about the outbreak in early January.

Madhav said that inconsistency in how different agencies report medical data can stymie algorithms. The text-scanning programs extract keywords from online text, but may fumble when organizations variously report new virus cases, cumulative virus cases, or new cases in a given time interval. The potential for confusion means there is almost always still a person involved in reviewing the data.

“There’s still a bit of human in the loop,” Madhav said.

Andrew Beam, a Harvard University epidemiologist, said that scanning online reports for key words can help reveal trends, but the accuracy depends on the quality of the data. He also notes that these techniques are not so novel.

“There is an art to intelligently scraping web sites,” Beam said. “But it’s also Google’s core technology since the 1990s.”

Google itself started its own Flu Trends service to detect outbreaks in 2008 by looking for patterns in search queries about flu symptoms. Experts criticized it for overestimating flu prevalence. Google shut down the website in 2015 and handed its technology to nonprofit organizations such as HealthMap to use Google data to build their own models.

Google is now working with Brownstein’s team on a similar web-based approach for tracking the geographical spread of the tick-borne Lyme disease.

Scientists are also using big data to model possible routes of early disease transmission.