I began systematically engaging with AI in 2023. Before that, my understanding of AI was largely confined to conceptual discussions or scattered application scenarios, with limited overall practicality. At least from the perspective of ordinary users and widespread industry adoption, AI truly entered the "usable stage" around the time of ChatGPT 3.0.

Over the past two years, AI's development has been evident to all. From initially being more experimental and demonstrative to gradually integrating into real production processes—participating in specific tasks like writing, programming, design, search, customer service, and risk control—AI's role has undergone significant changes. It may not perform perfectly in every aspect, but it can no longer be easily dismissed as a "gimmick" or "toy."

It is against this backdrop that I wrote my previous article, " Why American Lawyers Haven't Been Defeated by AI ," and engaged in an in-depth discussion with a lawyer friend on the topic. Additionally, I was invited to answer a question on a Q&A platform: "Why do many lawyers not acknowledge that law is one of the industries most impacted by AI?" This also sparked considerable discussion. This article is primarily based on my views on the issue and the related discussions.

Different Starting Points of Observation

Overall, I believe the question of AI's impact on the legal profession is not merely about assessing technological capabilities. It is more akin to offering entirely different interpretations of the same industry under different frameworks of understanding.

In related discussions, a notable phenomenon is that the parties involved are not always discussing the same thing.

Many lawyers base their judgments on specific practical experience. Real-world cases are not neat and tidy; clients' actions often lack rationality, communication is fraught with uncertainty, and there is significant room for non-standardized operations within procedures. Legal work is not as simple as "inputting facts to match laws and outputting conclusions." It involves experiential judgment, communication strategies, and an understanding of specific individuals.

Within this experiential framework, AI is often seen as a tool with limited capabilities. It can assist with retrieving information and organizing texts, but it struggles to truly understand the ambiguous, nuanced, and non-formalizable aspects of the real world. It is even less capable of assuming the judgment responsibilities borne by humans.

However, if we shift perspectives and view the legal profession as a large-scale social governance service rather than a highly personalized craft, the conclusion often changes.

From this viewpoint, legal work is inherently highly rule-based, text-intensive, and repetitive. A significant portion of the work revolves around organizing facts, summarizing evidence, applying rules, and advancing procedures—all of which can be deconstructed, standardized, and scaled. The discussion thus shifts from "Can AI adjudicate like a human?" to "Which aspects are already being reshaped?"

AI's Entry into the Legal Profession Often Begins with Processes

In many discussions about AI and law, the most frequently raised question is "Can AI deliver correct judgments?" However, in reality, AI's entry into the legal profession does not start with the most critical and sensitive adjudication aspects.

More commonly, AI first enters the highly repetitive, pre-procedural tasks that have long relied on human labor. For example, courts use OCR to process large volumes of paper documents during material intake, deconstruct factual elements in complaints into structured information using tables to standardize formats, quickly compare similar cases within trial management and supervision systems, and conduct preliminary analysis and labeling of images, videos, and audio recordings during trials and evidence review. These tasks do not directly determine outcomes but significantly alter how humans handle cases and improve efficiency.

Such changes are subtle but have cumulative effects. Once the methods for organizing and filtering basic information change, the cost, pace, and focus of subsequent judgments also adjust accordingly. From this perspective, the transformation of the legal profession resembles a gradual process of procedural restructuring rather than a one-time replacement.

What Is the Rule of Law, Exactly?

In discussions, one phrase is frequently mentioned by legal practitioners: "This is how things operate in reality today."

This statement is not incorrect, but it implies an assumption that the current state of the rule of law is a long-term, reasonable existence. When inconsistencies, disparities, or even chaos in the implementation of laws are seen as inevitable norms, any effort to enhance consistency and predictability through technology is deemed unrealistic.

However, if law is understood as a large-scale institutional arrangement aimed at uniform application and reducing randomness, then the imperfections in reality appear more as conditions that need gradual correction. In this sense, the disagreement does not entirely stem from differing assessments of technological capabilities but from varying understandings of the goals of the rule of law itself.

Don't Mistake Changes in Tools for the Disappearance of a Profession

I often draw parallels between the Industrial Revolution, where industrialization largely replaced handicrafts, and the development of AI.

However, in many discussions, people tend to assume that AI's involvement directly translates to "lawyers will be replaced." This concern is not entirely unfounded, but historical experience shows that technological progress more often leads to changes in work methods rather than the disappearance of professions.

Industrialization did not eliminate laborers; it transformed the organization of labor. Inefficient, non-replicable production methods were phased out, but people continued to find new roles within the system. A similar logic is likely to unfold in the legal profession. Basic tasks will become more standardized, repetitive labor will be absorbed by tools, and human value will increasingly manifest in discretion, handling exceptions, and assuming ultimate responsibility.

AI Is Already Deeply Involved in Legal Work

Returning to the initial question: many lawyers may not agree that "law is one of the industries most impacted by AI," not necessarily because they overlook the technology itself, but because the operational methods they are familiar with and rely on remain effective in reality.

However, technological influence often does not manifest by overturning existing systems. Instead, it gradually changes procedural structures by continuously intervening in peripheral aspects. When changes accumulate to a certain extent, the operational logic of the industry adjusts accordingly.

Ultimately, whether the legal profession will be transformed by AI is not a simple "yes or no" question. It is more about the pace, levels, and pathways of change. And this question is clearly already unfolding.