[In-Depth Look] Challenges in the Era of 'AI Human'

From the 5th to the 8th of last month, at the Consumer Electronics Show (CES) held in Las Vegas, USA, the domestic artificial intelligence (AI) specialist company DeepBrain showcased an AI human (virtual human) solution based on conversational AI. The AI human modeled after a female anchor from a broadcasting company spoke various languages including Korean, English, Chinese, and Japanese, with natural language and facial expressions. Another AI specialist company, MindsLab, collaborated with Shinhan Bank to realize the future appearance of bank branches. This company has already provided AI Contact Centers (AICC) to domestic insurance and card sectors. One domestic card company uses this company’s AI human for training recruiters. Additionally, the company has systems where AI humans can replace roles such as copyright monitors, wildfire monitors, process managers, and controllers. In March last year, the girl group of AI startup Pulse9 surpassed 2 million views on their music video on YouTube just five months after their debut.

The era of AI humans has truly arrived. It won’t be long before AI humans take over work in various fields. If combined with robotics technology, there may be very few areas left where humans must work directly. Positively, this can be seen as the long-held human ideal of liberation from labor being achieved. On the other hand, if humans liberated from labor are excluded from income-generating activities and cannot obtain the capital necessary for living, this would be a dystopia. This is the background for the introduction of basic income. There are also views advocating for significantly reducing traditional working hours, such as a four-day or three-day workweek without salary cuts, so that members of society can maintain a basic standard of living.

Meanwhile, in modern society where all economic activities are conducted through virtual spaces, the emergence of AI humans indistinguishable from real people inevitably raises concerns about misuse. AI humans that are more refined than the original person can hide and embellish the true self, potentially advancing voice phishing and internet phishing. AI humans that are more idealized than the original person may disrupt human relationships. In fact, in the recent election, the People Power Party used a deepfake-modeled ‘AI Yoon Seok-yeol’ (deepfake video) of their candidate Yoon Seok-yeol to interact with voters. By removing the candidate’s unnatural movements and irritating speech patterns, it was perceived as much friendlier by voters, sparking controversy over whether this violated the Public Official Election Act. Regarding this, the National Election Commission stated around the 11th that while election campaigning using deepfake videos is allowed if done within legal limits, failing to indicate that the video is a deepfake violates Article 250 (False Information Publication) of the Public Official Election Act. Conversely, if the deepfake video is clearly labeled as such, it is not considered a violation in principle. Deepfake refers to a production technique that uses AI technology to synthesize videos of a specific person’s face, creating images different from the original.

The new social trends based on scientific and technological advances such as AI humans cannot be stopped. The task is to establish laws and systems to prevent the misuse of science and technology, and above all, to fairly distribute the social benefits gained from technological progress to those members of society who might lose opportunities because of it.

Baek Ju-seon, Chief Attorney, Law Firm Yungpyeong