(image: Korea Bizwire)
SEOUL, Oct. 12 (Korea Bizwire) — An artificial intelligence (AI)-based service offered by a major web portal is now embroiled in controversy over fairness.
The Fair Trade Commission recently ordered Naver Corp., the nation’s biggest search engine, to pay a 26.7 billion won (US$22.9 million) fine for manipulating search algorithms for shopping and videos to gain leverage.
South Korean internet giant Kakao Corp. is also facing accusations that Kakao T, a cab-hailing app, prioritizes taxis owned by the company or its affiliates when allocating incoming hails.
When such controversy arises, these businesses claim that it is simply the result of AI decisions, which human intent could not possibly interfere with.
Experts, on the other hand, argue that AI algorithms that are currently being used are far from being technologically independent from human stereotypes or partiality since they are designed by humans and the data used for machine learning might have been biased.
“When we assume that the AI learns from multiple resumes to find the right person for a certain job, and it turns out that the majority of the resumes happen to be submitted by male applicants, the AI may inadvertently assume men are more suitable for the position,” said Prof. Yoon Jong-yung from Kookmin University.
“The AI can only learn from public data records that are mostly biased in their own ways, and humans are the ones responsible for developing AI algorithms, which makes it inevitable that our biased thoughts are reflected in the AI,” said Jeon Chang-bae, chairman of the board of the Korea Artificial Intelligence Ethics Association.
In the United States and the United Kingdom, both leaders in the AI world, AI algorithms are being widely used in both the public and private sectors, and have more often than not ended up in social controversy.
South Korea, too, is called upon by experts to engage in intensive debate over AI fairness and other related ethical problems.
“It seems like AlphaGo beating the world’s best Go player has instilled society with this impression that AI is somehow better than humans,” said Song Kyung-jae, a sociology professor at Kyunghee University.
“Now is the time for us to question that notion, and verify if AI algorithms really do work as intended.”
H. M. Kang (hmkang@koreabizwire.com)
