Walk into any standardized law firm, whether in Shanghai, New York, Mumbai, or Tokyo, and you will certainly see two types of lawyers: one is the associate lawyer, and the other is a former associate lawyer.

A doctoral degree in humanities and science means qualification to explore frontiers and指点江山 (make strategic remarks), while a doctoral degree in law, whether a Chinese Juris Doctor or an American JD, must go through the associate stage before becoming a lawyer.

Even the most famous former U.S. President Lincoln, as smart as he was and able to study law on his own, still had to spend a few years as an associate under an experienced lawyer before he could practice independently—a stage he could not bypass.

When names are not proper, words cannot proceed smoothly. The lack of an appropriate title for associates in China’s lawyer industry is not an occasional oversight but a systematic blind spot.

1. “Lawyer Assistant” Is Not Necessarily “Associate Lawyer”

Contract texts for venture capital transactions often undergo multiple rounds of negotiation and revision by multiple parties. Text checking and proofreading before final printing and signing is an unavoidable work环节. Should this task be undertaken by a lawyer assistant or an associate lawyer?

This is not a brain teaser. However, the answer is not so obvious:

a. In urgent battlefield conditions, even a cook can go out to fight with bayonets—regardless of titles?

b. If time is ample, resources are abundant, and management is orderly, it should be the “lawyer assistant” assisting the “associate lawyer” in executing this task.

a conforms to common sense and needs no elaboration. b is somewhat being clever, highlighting the difference between associate lawyer and lawyer assistant.

In fact,

Many law firms do not distinguish between associate lawyers and lawyer assistants, calling them all “lawyer assistants” in general. Few law firms use the title “associate lawyer.”

Associates aspiring to become lawyers often do not care whether their title has “assistant” first (left) or “lawyer” first (left)—they will sooner or later become “former” assistants. For those who cannot or do not plan to become lawyers due to educational and qualification limitations, they care even less about the textual structure of such titles.

A lawyer’s assistant can completely avoid involving legal affairs, just like assistants accompanying celebrity actors do not learn performance; graduate students handling their mentors’ expense reimbursement procedures become experts at attaching invoices, but this has nothing to do with academic pursuits. A lawyer specializing in medical dispute resolution may need a dedicated assistant to manage client (patient) medical record archives, while a lawyer engaged in trademark rights protection may have an assistant who day after day specializes in handling notarization procedures for screenshots—their work, while serving lawyers, does not itself involve lawyer thinking.

An associate lawyer is a lawyer with lawyer qualifications and necessary education,

just not yet able to independently make judgments on legal affairs for clients’ rights and interests.

An associate lawyer is a future lawyer.

Unfortunately, many associate lawyers get bogged down in lawyer assistant tasks, unable to feel the use of their legal knowledge. Some associate lawyers, due to the repetitive nature of similar assistant tasks, cannot feel intellectual enjoyment, which instead dampens their enthusiasm for lawyer work.

2. What Do Associate Lawyers Learn?

The conventional answer is “learn practical skills.” How to learn? The conventional answer is “the master introduces you to the door, but cultivation depends on the individual.” These are lazy people’s responses. The latter mystifies the learning mechanism and shifts the responsibility of guidance from instructors, while the former is actually answering a different question, because the original intent was: what to learn in practice?

All law schools teach contract law, but no law school has ever systematically trained how to review contracts. Remembering the essentials of contract law and reviewing contracts are entirely different intellectual activity processes. Reviewing contracts requires mobilizing common sense, and related skills can only be obtained in practice—the so-called “learning to swim by swimming.”

Let me give a practical example. A client requires completing the review of 10 Mutual Non-Disclosure Agreements (NDA) in a short time. The supervising lawyer is too busy to specifically guide the new associate lawyer on how to effectively complete the review. In other words, if the supervising lawyer had time to explain how to review, they might as well complete the review personally in about the same time. In the rush, the supervising lawyer guided the associate lawyer thus:

These 10 documents appear different but are actually the same type of documents in the same industry. Although 10 is not as many as 100, it is still enough to reflect what are standard clauses and standard expressions, and what are unorthodox clauses or non-mainstream wording. Viewing these 10 documents from this perspective, the associate lawyer actually has sufficient self-study materials to quickly become an “expert” on mutual NDAs. In other words, as a commonly used agreement, the basic logic and expression norms of Mutual NDA can emerge through 10 templates. The mutual NDA itself does not involve the buying or selling of specific products or service delivery, but rather governs how parties bear confidentiality obligations, so it does not require specific transaction backgrounds—the focus of review is on whether it is reciprocal, reasonable, and effective.

Such guidance gave the associate lawyer a grasp on an otherwise formless task. Rather than reviewing the 10 documents one by one in isolation, the associate lawyer discovers patterns and highlights models by comparing them with each other. With patterns in mind, the associate lawyer’s review no longer faces thousands of words piled up, but rather the clear structural relationships between clauses. Based on this, the associate lawyer can quickly form judgments and make annotations and suggestions.

In such a small practical project, the associate lawyer learned the core logic and standard expressions of NDAs, but this was not the most important gain. The gain that truly has long-lasting impact is that the associate lawyer has had a vivid experience of identifying patterns, mastering rules, and applying them to make judgments.

3. Artificial Intelligence: Inserting Itself Between Lawyers and Assistants

The associate lawyer mentioned earlier who could identify patterns, master routines, and self-study problem-solving based on 10 similar documents is because the associate lawyer is a highly educated and high-IQ professional. If it were a complete novice, could they also reach the same level through self-study training based on 100,000 similar documents?

Today’s fashionable Artificial Intelligence (AI), especially experts in the so-called neural network “Deep Learning” field, believe that there is no intelligent behavior that machines cannot learn, as long as there is a sufficiently large training database. If 100,000 is not enough, then 1,000,000.

AI technology is already changing all aspects of production and life. The legal industry is also feeling the benefits brought by AI. The quality of some machine-translated legal documents is astonishing, not to mention the speed and low cost.

The application of AI is not limited to translating legal documents.

On August 18 this year, Zhou Qiang, President of the Supreme People’s Court, mentioned at an international conference that China’s judicial system is “strengthening the integration of AI with trial and execution work… achieving automated handling of auxiliary affairs, intelligent case profiling, automatic evidence verification, precise push of legal provisions and similar cases, continuously improving judicial quality and efficiency.” Yang Yanchao, Director of the Science, Technology and Law Research Center at the Law Institute of the Chinese Academy of Social Sciences, said that his team [1] “conducted numerous experiments on a robot called ‘FILE,’ and ultimately it can write professional attorney briefs in the intellectual property field for lawyers.”

Peng Jiahao, Director of the Digital Governance Center at the Shanghai Artificial Intelligence Institute, envisioned [2]:

Taking GPT-3, OpenAI’s most advanced general language model, as an example, its parameter scale reaches an astonishing 175 billion, possessing powerful computing capability to complete complex tasks like translation, Q&A, and text generation—its generated news is indistinguishable from human works, even surpassing human levels. Whether we worry or not, AI will certainly produce results extremely close to those of judges in the future, transforming from an auxiliary tool to a partner working alongside judges.

A Lin Sister fell from the heavens—AI emerged between assistants and lawyers.

Is it a blessing or a curse? The cost-effective AI replacing some associate duties and eliminating manual review at the filing division is great; but AI grabbing the jobs of lawyers and judges? Hold on.

4. AI: The Associate Who Can Never Complete Apprenticeship

Like anything fashionable, AI systems are often “superficial and flighty despite showy appearances.” Ren Zhengfei said quite fairly and politely in a Face-to-Face interview on CCTV [3]: AI is just statistics!

The so-called “deep learning” is nothing more than adding multiple layers of neural network nodes between input and output ends, making the error correction mechanism between the system’s information input and output more detailed, thereby improving learning accuracy. The so-called “neural network” does not truly replicate or simulate human brain neural networks in any substantive sense—it merely makes a rough analogy based on the network form of the human brain’s nervous system. Many laypeople are “confused but awed” by these two mystical terms and dare not investigate deeply.

Compared to the deep-learning AI system that is “superficial and flighty despite showy appearances,” a lawyer’s self-study efficiency does not rely on repeated stimulation from massive amounts of similar data, because the essence of lawyer thinking is precisely discovering and applying new categories. Like poets, lawyers discover the different in what ordinary people see as the same, and discover the same in what ordinary people think is different. U.S. Supreme Court Justice Potter Stewart said, I cannot define “obscenity,” but I know it when I see it. Obviously, the justice does not need to see massive amounts of obscene works to form judgment. A poet said, “The night gives me black eyes, but I use them to seek light.” AI may mimic similar sentences based on statistical patterns of massive sample texts, but it is boring.

The tastelessness of AI writing poetry and the inadequacy of AI writing legal documents and judicial judgments both stem from today’s AI deep learning judging success or failure by results, completely ignoring that poetry and legal practice both have situations first, then results, and that right and wrong of results are always relative to specific situations, while situations depend on human intentions. AI has no and cannot have autonomous intentions—only assigned tasks. The data labels used to train deep-learning robots are set by humans. The labels themselves already embody human intentions.

What are situations and what are intentions?

When Shanghai was just liberated, Su Yu, as commander of the East China Field Army, and his wife Chu Qing were strolling down the street in a rare moment of leisure and encountered a café [4]. Su Yu said this place is nice. Chu Qing, born into a prestigious family, naturally knew the romantic and leisurely meaning of a café. Before she could happily pull Su Yu to walk in, Su Yu continued: the rooftop here can set up machine guns to封锁 (blockade) the entire area. The input physical scene was the same, but with different intentions, the situation was different, and the output reactions were completely different—no right or wrong, it just inadvertently became a historical anecdote.

A lawyer’s thinking is always rooted in specific situations, sensitive to the intentions of clients, counterparties, and relevant third parties, proactively shaping situations based on intentions, and discovering the most favorable ones as major premises among the various rules.

The remaining syllogistic reasoning is a simple task that associate lawyers or sufficiently powerful AI can execute.

Do you remember the IBM robot Watson, which made a splash a few years ago? It could win knowledge contests and defeat chess masters. But these so-called intellectual behaviors are worlds apart from lawyer thinking. Watson lacks common sense, cannot recognize and grasp situations, and can only do simple customer service Q&A within fixed parameter ranges.

Immature poets imitate, mature poets plagiarize. (This statement comes from T.S. Eliot; Steve Jobs made a similar statement in 1996: “good artists copy; great artists steal—and we have always been shameless about stealing great ideas.”)

The so-called plagiarism is borrowing. Why Eliot, Picasso, and Jobs were not ashamed of borrowing is because what was publicly borrowed was only form, but the situations to which they were grafted were vastly different, and the meanings conveyed were naturally different. Same form, different spirit—borrowing creates new meaning and thus constitutes creation.

Both associate lawyers and AI imitate. AI will only ever imitate, while associates, based on imitation, will try autonomous borrowing and grafting in new situations, making connections and thus becoming independent lawyers.

Notes:

[1] http://iolaw.cssn.cn/bwsf/202008/t20200805_5166326.shtml

[2] https://www.thepaper.cn/newsDetail_forward_18756992

[3] https://cloud.tencent.com/developer/article/1400717

[4] https://new.qq.com/rain/a/20200705A0CEYG00