The two faces of generative AI: Industrial innovation or risk without a safety net? > News

Recently, domestic and foreign industries are using generative AI as a key driver of business innovation. Following Samsung Electronics' introduction of ChatGPT at the company-wide level, SK Hynix also reviewed its security system and formalized the use of externally generated AI. This is not simply following trends, but a strategic choice to secure competitiveness in the global market by maximizing the efficiency of the research and development (R&D) process and speeding up decision-making. In fact, Samsung Electronics confirmed through internal testing that ChatGPT has the highest business suitability and user satisfaction, and is linking this to the group-wide AI conversion (AX) strategy. Companies are now choosing technologies beyond the performance of their AI models, based on how well they work in real-world situations and accurately capture the intent of the question being asked.

However, criticism is pouring in that the safety measures that support it are still weak compared to the speed with which the technology is spreading. Attorneys general from 12 states in the U.S. issued subpoenas to OpenAI, taking aim at ChatGPT's safety issues. This is a warning that we can no longer ignore the social side effects that AI can cause, such as its connection to crime, issues with protecting minors, and answers that increase users' psychological anxiety. In fact, Open AI has become at the center of legal and ethical responsibility as suspicions have been raised that the shooting suspect used AI to plan the crime or that AI's responses influenced the extreme choices of certain people. From the perspective of Open AI, which is preparing for an initial public offering (IPO), these regulatory authorities' investigations are acting as a huge variable that increases uncertainty in the listing process.

Even for individual users, AI acts as a double-edged sword. Recently, a number of cases of unauthorized payment of large amounts of money have been made due to the theft of card information from people who have never used ChatGPT, revealing a hole in the security system. This is more than just a technical error, it is a serious security incident that fundamentally threatens the reliability of the platform. Also, as in the case of Japan, as the number of young people choosing AI instead of people as a counseling center increases, the limitations of AI's advice and issues of responsibility have also surfaced. Although AI has outstanding intelligence to point out details that humans miss, we must never forget that it does not bear any responsibility for the results of its answers.

Nevertheless, the potential of generative AI still functions as a warm technology that breaks down social barriers. The Busan International Council for the Disabled published a collection of sayings with illustrations created by people with disabilities using AI, showing an example of the digitally underprivileged class becoming confident creators beyond consumers of technology. This proves the value of AI as a technological welfare and suggests the right direction in which we should move. Meanwhile, in the Korean market, fierce competition between services continues, with Antropic's 'Claude' shaking up the two-tier system of ChatGPT and Gemini and jumping to second place in sales. This market dynamic provides users with better options, but at the same time, it suggests that the cautious approach and information protection awareness of users who utilize various models in parallel are becoming more important.