[Column] The ChatGPT shock

Kim Dong-ho

The author is the editor of economic news for the JoongAng Ilbo.

After all the sensationalism surrounding ChatGPT, I downloaded its app on my smartphone. When I typed in a couple of questions, I was awed by the level of intelligence of the chatbot developed by OpenAI, a San Francisco-based research company. First of all, I could talk with AI in real time. The chatbot completed plausible sentences immediately and smoothly. My questions could not frustrate its intelligence.

Even the shock from AlphaGo seven years ago cannot be compared to ChatGPT. In March 2016, the AI Go player repeatedly beat its lofty human rival Lee Se-dol — now-retired but unrivalled at that time — in all but one of the five matches. Foreshowing the advent of the AI revolution, the first match between AI and a human Go master, held in Seoul, was broadcast live to the rest of the world through YouTube. Eighty million people around the globe watched the thrilling Go games holding their breath.

The start of ChatGPT was relatively quiet. But after Google and Miscrosoft entered the race for AI, subscribers to the chatbot have increased exponentially. It took 42 months for Netflix to reach 1 million subscribers, 30 months for Airbnb, 24 months for Twitter, and 10 months for Facebook. The fastest one — so far — was Instagram which needed 2.5 months to reach the milestone.

But ChatGPT accomplished that in just five days. Its subscribers surpassed 100 million in just two months. The biggest strength of the chatbot is its ability to organize — and streamline — a gargantuan collection of ideas and thoughts. Search engines can’t do that. Besides, any question is possible. For instance, if you ask, “Why was the Miracle on the Han River possible?” the chatbot summarizes its answer in three or four paragraphs without any redundancy.

You cannot get such lucid results from Google or Naver. But if you ask the chatbot to make a prediction for the 2023 semiconductor market, it gives you concise and logical answers by getting to the point. It takes a while for humans to give an answer on such levels. The chatbot is drastically different from simple chatbots used by banks or security firms, as it can answer any question beyond the boundaries of human professions.

AI is rapidly replacing human labor, from self-driving cars to self-ordering kiosks at restaurants. ChatGPT, which can self-produce new contents based on big data, is totally different from them, as it can make logical suggestions and proposals that only humans had been able to do. The remarkable development was possible due to the formative AI and high-performance chips.

Lawyers and doctors have started feeling the tension. Local lawyers in particular must confront a challenge from ChatGPT following an earlier one from LawTalk, an online legal counseling service operated by Law&Company in Korea. Lawyers are threatened by ChatGPT as their clients can gauge basic directions of any legal disputes from the chatbot without having to pay hefty legal fees. A number of high-paid professions also will face such a predicament.

ChatGPT also will make it harder for malicious YouTubers to spread fake news, because people can find, though roughly, the truth behind heated controversies through the chatbot. It is politicians who should be most alerted by ChatGPT, as you just have to ask the chatbot if they made irresponsible pledges or not.

Out of curiosity, I asked the chatbot what group of opinion leaders is the biggest obstacle to advancing Korea further. I advised it to choose among politicians, civil servants, attorneys, professors, corporate leaders and journalists. The chatbot immediately came up with this answer: “It is difficult to single out a particular group. But politicians can be the largest stumbling block in the sense that they fix directions and systems of a nation.” How could a chatbot be smart like this?

I was emboldened to ask a provocative question to the chatbot. “Won’t ChatGPT steal jobs from humans fast?” Its answer was rather a masterpiece. “ChatGPT is designed to automatize a certain task. Therefore, it will not take away jobs from humans, but can liberate them from labor by reducing their working hours.” The tone was gentle. But it sounded like a demand that humans hand over their simple office job to ChatGPT. Last week, Google fell from grace after its new AI chatbot “Bard” gave the wrong answer to a question about a telescope. But you cannot avoid the change. How much will ChatGPT — and its future competitors — change the world as know it?

Welcome to the brave new world!