Series Index
Part 1 | Attorney Hong Seung-kwon: “AI Enters the Legal Field”
Part 2 | How Should Lawyers Use AI?
By Hong Seung-kwon
Principal Attorney, PAN Law Firm (skh@law-pan.com)
Lawyers today tend to fall into two broad groups. Some actively use AI in their work, while others avoid it almost entirely. Many assume the first group will quickly become the majority, but the reality inside law firms feels slower and more uneven than people expect. Even now, it is easy to find lawyers who barely touch AI.
Out of curiosity, I once asked a colleague, “Why don’t you use AI?” The answer was blunt and surprisingly persuasive. “At first, I used it constantly,” my friend said, “but after a while, I couldn’t tell whether I knew something myself or whether the AI knew it. So I stopped.”
That comment stayed with me. We already live in a world where most information can be found through a simple search, yet the distance between “what I truly understand” and “what I can retrieve” remains enormous. Rather than treating AI as a shortcut, I decided to use it more deliberately—as a tool for self-checking, verification, and sharpening my own thinking.
The two AI tools I use most are Super Lawyer and ChatGPT. Both require paid subscriptions, and in practice I use them almost continuously. Still, the two tools play very different roles, and understanding that difference matters more than choosing the “best” one.
Super Lawyer is designed as a legal AI, built around its own structured database of case law and related materials. Because it relies on anchored legal sources, it is less prone to the kind of hallucinations that appear more frequently in general-purpose models. Another advantage is traceability: when reviewing statutes, doctrines, and legal principles, it is often easier to identify what authority the answer is based on.
I rely on Super Lawyer most heavily when I need to organize the key issues in dispute and quickly check whether similar cases have already been decided. It is especially useful when I am mapping the range of precedent trends and trying to confirm what kinds of reasoning courts have accepted under comparable facts. It can also provide an initial sense of which legal provisions are likely to be relevant in a legal environment where the number of active laws and regulations is vast and constantly evolving.
ChatGPT, by contrast, is a general-purpose language model trained on large-scale text. Its strength is breadth: it can discuss almost any topic in fluent, persuasive language, and it responds instantly. At the same time, its weakness is also clear—accuracy is not guaranteed, and confidence does not always equal correctness.
If I ask ChatGPT about a legal issue, it often provides a clean, immediate answer. But there are many occasions where the cases, statutory provisions, or legal principles it cites do not actually exist, or where the source cannot be verified. Sometimes a small change in how a question is framed produces a completely different conclusion, and as issues become more complex, it can miss the core dispute or misread crucial facts.
Even so, I do not consider ChatGPT useless—far from it. Because it draws from an enormous amount of text, it can be valuable when I am trying to explore possible directions for solving a legal problem. More than once, it has offered a clue when I was stuck and had no clear way to approach the issue.
Based on my experience so far, AI tools are undeniably useful, and they are likely to become even more helpful as they improve. But their limitations are just as visible. No matter how effectively an AI finds related precedents or organizes statutes and doctrines, deciding a real case operates on an entirely different level.
Legal judgment is ultimately the process of finding an answer that is meaningfully workable for the people involved. That requires attention to context, emotion, timing, and relationship dynamics—elements that rarely fit neatly into written facts or legal structure. There are aspects of a dispute that cannot be captured through analysis alone, because what matters is not only what happened, but what it means to the parties now.
In the next article, I will discuss what “judgment” truly means for a lawyer, using specific cases I have personally encountered while working with AI.
