AI is not a threat to lawyers. She threatens their reasoning,... if we let her do it

Descartes said in 1637: “I think, therefore I am.”

It assumes that our representations and perceptions are sources of error. It is impossible to build a correct reasoning, to access any definitive truth. Two ways out of the impasse according to Descartes: doubt and the obsessive search for the truth.

· Doubting every idea that comes to mind takes us away from error and delusion. This is when the first stable truth appears: 'I think therefore I am' in the sense of 'I doubt therefore I am'.

· The obsessive search for the truth. Every certainty must be able to be questioned. Each demonstration must be able to be refuted. Each theory must be able to be overcome.

This human capacity is that of thinking for himself. It is the desire to gain autonomy in relation to the physical world and to one's own mental mechanisms. We are at the heart of the critical spirit that is talked about so much today in executive circles.

Four centuries later, we are surrounded by machines that “think,” or at least produce something that resembles thought enough to trouble us. They qualify, interpret, weigh, and write.

For the world of legal professions, this is a historic and existential moment. Because what makes a lawyer is Jurist is his ability to categorize, to interpret, to discern, to carry values to organize human relationships. For a corporate lawyer, it is this same cognitive tool that allows him to organize relationships between stakeholders.

So an urgent question arises: at a time when many legal departments are equipping themselves with AI, how can we integrate AI uses without sacrificing legal reasoning? Better: how to integrate AI to magnify it?

The last Grenelle du Droit asked: How to train in AI without sacrificing legal critical thinking? I have had numerous conversations with young lawyers and students and the fact is that a mixed feeling dominates: excitement and melancholy. An excitement to experience a period of unprecedented innovation and a form of melancholy when thinking about skills that would be obsolete even before having exercised them.

In short, for some, the Cogito has transformed into Seum.

However, I don't believe this story.

This paper explores 3 aspects in succession before trying to reach a conclusion: 1. Understand the nature of legal reasoning analyzed through the skills mobilized. 2. Explore the apparent opportunities for hybridizing these skills with AI as well as its points of vigilance 3. Take into account the new risks of this integration brought about by the most recent studies and experiments

Legal Reasoning Skills

There are several exercises that implement legal reasoning (legal analysis, judgment commentary, etc.). The idea here is to identify the principles that underlie all the routines, practices and activities of lawyers. In other words, what are the cognitive skills that every lawyer uses every day in his work: Legal Reasoning Skills

The analysis excludes other skills necessary for the lawyer such as specialized legal, behavioral, business or managerial skills.

The 6 principles of reasoning

Legal reasoning is not a homogeneous block. Historically, it is based on six principles. What are they?

The legal syllogism. The principle is to articulate the major (rule of law) with the minor (facts of the case) in order to draw a legal conclusion. It is the backbone of reasoning in continental law.

Skills mobilized:

· Deductive logic: know how to properly link the three times.

· Editorial rigor: clearly formulate the rule, the relevant facts, and then the conclusion.

· Structuring ability: organize a text as a demonstration, not as a story.

The legal qualification of the facts. The principle is to transform “raw” facts into legal categories: e.g. “He loaned money to his friend” → loan contract. “He pushed someone down the stairs” → misdemeanor, violence, etc.

Skills mobilized:

· Legal culture: know the main categories (contract, responsibility, property, procedure...).

· Abstraction: see the legal “form” behind the facts.

· Attention to detail: an apparently trivial fact can change the qualification.

The hierarchy and consistency of norms. The principle is to remember that all rules are not equal: Constitution > law > law > regulation > contract, etc. The lawyer must choose the relevant norm and check that it is not contradicted by a higher standard.

Skills mobilized:

· Knowledge of the sources of law: texts, case law, doctrine.

· Ability to “map”: locate each standard in the global architecture.

· Critical vigilance: identify conflicts of norms, exceptions, derogations.

The interpretation of texts and precedents. The principle is that a text never “speaks” by itself. It must be interpreted: i) literally (meaning of words), ii) by the purpose of the rule (the “spirit” of the rule of law), iii) systematically (in coherence with the rest of the law)

Skills mobilized:

· Fine reading: text, decisions, preparatory work

· Historical and teleological culture: understanding the “why” of the rule

· Argumentation: justify why one interpretation is preferable to another

Reasoning by analogy and by distinction. The principle is based on the use of analogy: “This situation resembles a case already judged → we apply the same solution”. Symmetrically, the reasoning identifies the distinction where appropriate: “This situation is different on this key point that justifies a different solution”

Skills mobilized:

· Recognition of Patterns : identify what is really similar or different

· Argumentative finesse: show that the difference is legally relevant (or on the contrary negligible)

· Capacity for legal storytelling: telling the facts in a way that approximates or distances a precedent.

Balancing interests and principles. When several rights or principles conflict (freedom of expression vs. right to privacy, security vs. freedom, etc.) , the judge or legal adviser cannot simply “apply” a rule: they must balance.

Skills mobilized:

· Systems thinking: see the effects of a solution on the entire legal order.

· Political and ethical sense: understand the values at play behind the norms.

· Ability to weigh: argue why, here and now, a certain principle should prevail.

These six principles are the daily lifeblood of teams of lawyers regardless of their specialty (contract law, M&A law, etc.). What irreducible body of skills does this represent?

Fundamental skills in dialogue with AI

It is possible to cross the large blocks of legal reasoning necessary with 10 fundamental skills.

These skills correspond to the core taught in law programs around the world. In France, they are also those that professional competitions systematically evaluate: CRFPA, ENM, INSP, A+ administrative competitions. Legal departments identify these skills as the most critical for decision-making, risk management, and internal credibility.

What dialogue can each of these skills maintain with AI?

1. Knowledge and prioritization of sources Of the law

- Principles concerned (major): legal syllogism and hierarchy and coherence of norms

- The opportunity for hybridization: the lawyer does his research with a unique interface. the AI connects to sources, restores them in natural language. As long as he is up to date, he knows the rules of hierarchy of norms.

- The risk: the lack of reliability: AI hallucinates by inventing case law judgments. The idea is then to focus your research in 100% controlled corpora (RAG, Open Data, MCP)

2. Pattern recognition 

- Principles concerned (major): reasoning by analogy and by distinction

- The opportunity for hybridization: AI excels in factual “clustering”. Its operation is based on neural networks that find semantic similarities in texts.

- The risk: gray areas may exist. Sources must be checked for completeness

3. Writing skills

- Principles concerned (major): syllogism, legal qualification of facts, interpretation of texts and precedents, reasoning by analogy and distinction

- The opportunity for hybridization: AI as a major language model excels at writing in various languages. Styles and formats can also be adapted

- The risk: the presence of errors or omissions, especially in successive summaries.

4. Deductive logic

- Principles concerned (major): the syllogism

- The opportunity for hybridization: The AI can verify that the reasoning is coherent, complete, non-circular.

- The risk: AI will not do a “real” verification of reasoning: it approaches a “canon of reasoning” by the mass of legal analyses with which it is fed.

5. Legal culture

- Principles concerned (major): legal qualification of facts, interpretation of texts and precedents

- The opportunity for hybridization: AI helps to contextualize by looking for facts in correspondence, by synthesizing relationship histories, by offering comparative analyses, etc. that humans do not necessarily have in mind. It serves as an “external memory”: the lawyer can focus on analysis rather than on memorization.

- The risk: the loss of historical and doctrinal depth. AI sometimes does work “on the surface”. Moreover, legal culture is not only a store of information: it is an industrial culture, a particular sensitivity to industry. (“in our sector, we never go to a “clash”)

6. Power of abstraction

- Principles concerned (major): legal qualification of facts, interpretation of texts and precedents, reasoning by analogy and distinction

- The opportunity for hybridization: AI stimulates abstraction. Its construction and training were based on the labeling of categories. For example, the lawyer can use the machine to list the legal categories in play. It can test several abstractions (“What if we qualified this situation differently? ”).

- The risk: the misuse of this machine capacity. Avoid using it to find complex and novel legal solutions. “the last mile must remain human”. Abstraction is a creative intellectual act, not statistical recombination.

7. Fine analysis ability

- Principles concerned (major): legal qualification of facts, hierarchy and coherence of norms, interpretation of texts and precedents, reasoning by analogy and distinction, balancing interests and principles.

- The opportunity for hybridization: AI detects micro - inconsistencies, conflicts of norms not seen. It accelerates the detection of weak signals

- The risk: the presence of misunderstandings on specific legal concepts for which the machine has not been trained. AI can provide a fine but erroneous analysis for this reason. Generalist models (LLM hyperscalers) expose lawyers to this risk

8. Systems thinking

- Principles concerned (major): interpretation of texts and precedents, balancing interests and principles.

- The opportunity for hybridization: AI maps dependencies (norms, risks, actors, processes). It simulates scenarios: “If you change this clause, what happens elsewhere?” The lawyer places weight in the decision-making criteria. He lets the machine calculate but ensures that the decision parameters are indeed his.

- The risk is linked to “interstitial” and unformalized knowledge that the machine ignores and that deserves to be entered into the system or collected by interviewing experts

9. Political/ethical sense

- Principles concerned (major): interpretation of texts and precedents, balancing interests and principles.

- The opportunity for hybridization: AI makes it possible to test several ethical frameworks (proportionality, equity, nondiscrimination, etc.). It helps to anticipate the reactions of stakeholders (customers, authorities, media, etc.). It highlights reputational risks by recalling, for example, similar precedents or crises.

- The risk is the “alignment” of the LLM if it is used indiscriminately: its underlying values (the alignment is its Political dressage pre-marketing) may conflict with the organization's value system/ethics

10. Capacity for discernment

- Principles concerned (major): Legal syllogism, legal qualification of facts, hierarchy and coherence of norms, interpretation of texts and precedents, reasoning by analogy and distinction, balancing interests and principles.

- The opportunity for hybridization: AI requires the lawyer to formulate what he considers relevant in the analysis and decision, which clarifies the thought. It offers alternative angles: AI can reveal interpretations or risks that the lawyer did not consider.

- The risk is low in theory

In summary, it is therefore appropriate to hybridize human skills linked to legal reasoning. It is possible to magnify it provided you master a fairly long list of points of vigilance.

Extensive potential for AI hybridization

In total, as we can see, the potential for hybridization is strong between the “bricks” of legal reasoning and the advantages of artificial intelligence.

However, as the period of decline in the use of AI in business has been getting longer since 2023, studies point to a new fact: the risk of cognitive atrophy from prolonged exposure to the use of AI.

Why do some skills atrophy

Recent studies tend to be more nuanced by highlighting the risks of prolonged delegation of certain skills to AI [“The Impact of AI and Legal Tech on Junior Lawyers: Risks, Opportunities, and Career Strategies” (2026) by Laurence Simons, and “AI and the Law: Navigating the New Legal Landscape” (2026) from Harvard Law School]

These studies identify endangered skills such as fine legal analysis, the ability to identify nuances, the construction of arguments, and the development of judgment. The major risk is the atrophy of basic skills if lawyers no longer manipulate the “raw materials” of law and beyond that, the loss of expertise of the organization in the long term if adoption is made massively and indiscriminately.

How do you explain it?

Not all legal skills react in the same way to technology delegation. The skills procedural (research, sorting, standardized writing) can be outsourced without lasting loss: they are based on stable routines. On the other hand, skills analytics(fine analysis, abstraction, profound legal culture) and especially skills reflexive (discernment, systemic thinking, political/ethical sense) are based on cognitive micro-gestures and internal deliberations that can only be maintained through practice.

If we take the table of legal reasoning skills and if we segment the skills according to the 3 procedural, analytical and reflexive families, the uses of AI are envisaged in a specific way

It appears that AI, by automating certain layers of this reasoning, requires us to revisit the balance of skills mobilized between delegation to AI and augmentation by AI:

1. For procedural skills, maximum delegation to AI is possible, the lawyer saves time without necessarily increasing his level of value (quality, customer satisfaction,...). This concerns knowledge and prioritization of sources of law, pattern recognition and writing skills. The challenge is productivity.

2. For analytical skills, the lawyer is enhanced by AI on the extended memory of facts and relationships, on categorizations and on fine controls. The lawyer takes a new but limited amount of time to use AI (increased working time) but provides greater value (in production, in responsiveness, in quality). The scope concerned here is deductive logic, first-level legal culture, the power of abstraction and the capacity for fine analysis. These skills must be maintained by humans to successfully communicate with AI. This dialogue is part of “sparring partnering”.

3. For reflective skills, the lawyer is enhanced by AI on complex dependency maps and on the simulation of decision models. Here too, the lawyer takes a new amount of time to use AI (weaker than for analytical skills because for more specific activities) but delivers even better performance (in production, in reactivity, in quality). In fact, reflexive activities carry strong strategic challenges (negotiation, litigation strategy, lobbying, etc.). This is a competence area with a very high value of systemic thinking, political sense and capacity for discernment. These skills need to be maintained through practice. The use of AI here is occasional but strategic, to be planned at key moments.

Towards a magnified Cogito?

Let's talk again to our imaginary lawyer at the beginning of this text who feels the “Seum”:

“If the machine thinks for us, if we abandon our highest intellectual gestures to it: discerning, interpreting, weighing, putting into perspective, then we will only reap a double impoverishment:

· A historical countersense, first of all: legal reasoning was never intended to produce mechanistic answers, but to affirm human freedom through deliberation. To think for one's fellows is to organize their coexistence. To delegate this is to renounce what the law is based on.

· A silent disappearance, then: that of the lawyer - an employee reduced to the rank of executor, deprived of his reflexive role, unable to articulate values, risks and strategy. A lawyer who no longer reasons is not “augmented.” He is a diminished lawyer.

It is now clear that there is no valid reason for this scenario to happen. But the story is being written and I see 4 conditions for success:

Adopt AI in a differentiated way according to the 3 families of skills : procedural skills can be delegated without the risk of cognitive atrophy on the one hand and with productivity gains on the other hand. Analytical and reflective skills can be increased by AI: not by delegating them but by increasing them. New ways of working with the machine must be considered, most often in sparring partner mode.

Experiment collectively with new ways of working. Get out of individual screens to validate the positive effects of delegation and augmentation by AI by following specific work protocols. Experiments conducted in A/B testing make it possible to demonstrate the superiority of one way of working over another. Alternate work with and without the machine on the same procedures. Is legal reasoning atrophied or magnified? And under what conditions is it magnified?

Ensure that the AI system is compatible with the new ambition : the “limits” are often linked to inadequate source data (not secure, not trained, not sovereign,...). Apart from the LLM or the LLM package, Data remains the number one subject to be dealt with in legal departments.

Investing new skills to assert the power of Corporate deliberation that corporate lawyers embody. In particular, invest in behavioral skills (influential communication, reasoned negotiation, advisory posture, etc.) and business culture.

And if you want to be a step ahead, Waldenlab can structure a discussion with you:

· Mapping the cognitive skills of teams: what opportunities? What threats?

· Define Jurists/Machine hybridization while mapping out legal reasoning

· Frame the data/AI infrastructure project to ensure that the data is reliable

The machine can coogitate. It is up to the lawyers to act.

‍