AI’s newest students

The rich have a developed nose to smell money. So, what have they been up to this summer?

The Korea Chamber of Commerce and Industry recently invited Jeong Jae-seung, professor at the department of Bio and Brain Engineering of the Korea Advanced Institute of Science (KAIST), to lecture. Jeong predicted artificial intelligence (AI) would become complete within the next two to three decades. He advised students to avoid fields like English and math where AI would take over. Instead humans should take leisurely walks or daydream to fall upon a “eureka moment,” he said.

Members of the Federation of Korean Industries, the lobby group for top chaebols, got a lesson on AI from Park Myung-soon, head of SK Telecom’s emerging technology research and development center. By 2020, revolutionary changes from AI will be obvious, he declared.

Samsung Electronics Vice Chairman Jay Y. Lee in June had some interesting questions for Oh Joon-ho, KAIST professor and expert in mechatronics and robotics. Oh is the inventor of the Korean humanoid robot Hubo. “When do you think AI will catch up with the human mind? If there is one thing God would want to change in humans, what would it be?”

In July, he was seen in a photo walking with Ginni Rometty, chairman and chief executive of IBM, during the annual Sun Valley Conference. Lee regularly attends the annual media finance conference, where he meets with Apple CEO Tim Cook and Larry Page, CEO of Alphabet, the holding company of Google.

A senior official at Samsung said the company was connected to IBM because of AI. Although Google’s AlphaGo has been in the media spotlight for its feats against human Go players, IBM’s Watson is the most advanced cognitive computer that has the industry’s attention.

IBM pulled out of the PC hardware business in 2005 and has invested heavily in the artificial intelligence sector. Watson has been active in finance, retail, education, and medical care. Its contribution to the medical field has been staggering.

In 2013, it delved into 600,000 types of medical evidence and studied 2 million pages sourced from a variety of medical journals to come up with the right answers for major hospitals to help treat patients. No human physician can learn that amount of information in a lifetime; Watson is now better than oncologists when it comes to diagnosis.

According to the American Cancer Society, the diagnostic accuracy rate for cancers by physicians was around 80 percent. For Watson, it was 98 percent for colorectal cancer, 91 percent for bladder cancer, 94 percent for pancreatic cancer, and 100 percent for cervical cancer.

While doctors correctly diagnosed lung cancer 50 percent of the time, the accuracy rate for the IBM supercomputer was 90 percent.

Diabetic retinopathy is a serious illness that can lead to loss of sight when it is not cured in early stages. The AI computer was as accurate as a veteran doctor, diagnosing it at the precise rate of 83 percent.

AI experts predict that in a decade, the only field left for human doctors will be psychiatry. Computers will soon replace humans in anesthesia, image medical science and pathology.

The Sedasys anesthesiology machine, an early adopter of AI, has brought down medical costs 90 percent. Sedation that used to cost $2,000 has been reduced to $150 to $200. Anesthesiologists who earn an average of $277,000 a year lobbied hard against the machine.

Sales were halted after a year. But some see an inevitable comeback as the machine can no longer be boycotted due to safety concerns, given that it has been tested under stricter guidelines than human doctors.

Time magazine pondered whether “Google Can Solve Death” in its most recent cover story. It posed the question not to hospitals and pharmaceutical companies but to Internet giant Google because of AI.

AI researchers are moving their nests to tech companies like Google, Facebook, Amazon and IBM. High compensation is not the only reason. They need supercomputing systems for AI research, which few university labs can afford. Google runs more than 1 million servers while Amazon and Facebook have fleets of more than 500,000.

Where is Korea in the race for the future? Kim Jin-hyung, Korea’s pioneer in AI research and an honorary professor at KAIST, sighs when asked. “AI must be prepared, not predicted.”

Over 300 engineering majors graduate from Stanford University every year, compared with 55 from Seoul National University. Computer software is part of the mandatory curriculum at elementary schools in the United States and Britain. There is no comparison, he said. It is a relief that at least corporate CEOs have an interest in the area.

*The author is an editorial writer of the JoongAng Ilbo.

Lee Chul-ho