[Editorial] Regulating ChatGPT

Korea has to start discussing issues emerging from the explosive growth of generative AI services

Tech buzzwords tend to come and go quickly. Remember the metaverse? Or the fourth industrial revolution, a concept that swept South Korea several years ago? Now, all attention is placed on what is called “generative AI,” led by OpenAI’s ChatGPT.

The popularity of artificial intelligence-powered ChatGPT, which generates highly human responses to text queries based on algorithms and a vast volume of data sets, is sparking off a whole host of new phenomena in various applications including computer coding, essay writing and information gathering. And a flood of AI-based uber-realistic photos and illustrations produced by carefully arranged text prompts is taking the internet by storm.

Some welcome the advent of extremely cheap and quickly produced useful answers as well as potentially commercial images based on advanced AI solutions. Others joke that a number of commercial industries and professionals may be doomed by the intensifying attacks of AI-generated solutions. A look at the fast-expanding adoption of generative AI suggests this dystopian prospect is not so far-fetched.

As with other tech fads, however, generative AI chatbots might lose their momentum at some point after their limitations are revealed and stringent regulatory measures are taken around the world.

On Friday, Italy became the first Western country temporarily to block access to ChatGPT over privacy issues. The main reason lies in the murky way OpenAI collects and processes data, some of which are scraped from internet forums without the consent of users. Not only the lack of transparency related to privacy but also the accuracy of information generated by ChatGPT has been under fire. Fake news, aided by super-realistic AI-generated photos of famous people, are already popping up in cyberspace.

After Italy’s move to ban ChatGPT, other European countries are reportedly reviewing wildly popular chatbot services to see whether strict regulations are needed. Authorities in France and Ireland contacted their counterparts in Italy to find out more about the ban. Germany is said to be considering blocking ChatGPT over data security concerns.

Privacy and other related issues are not limited to ChatGPT. Other generative AI services from Google and other major tech firms are also saddled with a slew of potentially damaging problems, including the massive exploitation of content available on the internet, copyright infringement and the biased manipulation of data.

South Korea’s major companies in the finance, IT and semiconductor fields are already taking actions to minimize the side effects of chatbot services. For instance, Samsung Electronics is reportedly conducting an internal survey on ChatGPT before drawing up guidelines about whether it can be officially used for work purposes. A growing number of local companies are rushing to set up rules since more cases in which confidential data got leaked are reported. Typing in confidential programming codes in ChatGPT, for example, can result in accidental data leaks.

Posco has opted for allowing its employees to use only its internal chatbot service to strengthen data protection. SK hynix has blocked the use of ChatGPT in its intranet, while allowing for limited use only when formal permission is granted following a security review.

But banning and restricting the use of generative AI services could bring about potential problems, as well. If too many local regulations are imposed on new advanced chatbots, the fledgling AI industry might be stuck, while other countries are encouraging AI tech firms to advance related solutions.

The fear of being left behind is real. Italy's Deputy Prime Minister and Infrastructure Minister Matteo Salvini said Tuesday that national competitiveness will suffer compared with other European nations if ChatGPT is not reactivated soon. Given the dizzy pace of advances in AI, such concerns are not groundless.

The Korean government has to study generative AI seriously and start exploring its positive and negative impact on the business, academic and science fields, among others. More importantly, it should make efforts to come up with balanced guidelines that aim to protect users’ content from reckless data exploitation by chatbot providers and also encourage businesses and individuals to utilize chatbots safely for creative applications.

Policymakers should be aware that as far as AI is concerned, timing is crucial, and they don’t need to ask ChatGPT about when they should start -- because the answer is most certainly “right now.”

By Korea Herald(khnews@heraldcorp.com)