AI is everywhere. From social media algorithms and smart thermostats to word processing. Almost everyone uses it to some extent, and it helps us work faster and more efficiently. At least, that is our view of the use of AI, but is this view correct? And what about lawyers? That is precisely what Cees Zweistra, assistant professor of law, ethics, and technology, and Julie Hoppenbrouwers, PhD candidate in disruptive technologies in legal practice, investigated for their new book Sneller & Beter? Het vraagstuk van technologie en AI in de rechtspraktijk (Faster & Better? The issue of technology and AI in legal practice). In this article, they discuss the impact of AI on legal practice and why lawyers should take a critical look at the use of AI.
"Preliminary, as yet unpublished research conducted by the Center for Law, AI & Design in 2025 shows that the majority of lawyers surveyed (>90%) use (legal) generative AI systems such as CoPilot, Harvey, and CoCounsel. These tools are used for tasks such as structuring ideas, generating summaries, translations, and document improvements," say Zweistra and Hoppenbrouwers.
"The use of generative AI, specifically large language models (LLMs) for knowledge production, seems to be becoming increasingly dominant in legal practice. Such models make it possible to 'outsource' the interpretation of the law. In the long term, this could have significant implications for legal practice," says Hoppenbrouwers. Zweistra emphasizes the potential consequences of this development: "Providers of legal AI systems will gain a more dominant position within the law, legal practice, and even have an impact on the formation of the law. This raises questions about the future independence of actors in legal practice and how legal practice can ensure that the interests of AI providers and new AI tools are in line with the interests of legal practice. After all, independence is a crucial pillar of legal practice, and that pillar is under pressure."
"We see that legal practice is looking for possible applications of AI, rather than possible problems for which AI could offer a solution. This needs to be reversed: the problem must be central, after which a suitable solution must be sought," says Hoppenbrouwers. AI is seen as a means of making lawyers more productive. "But if you look at our history with IT applications such as e-discovery (a tool for searching digital documents for evidence in criminal proceedings), it turns out that this has not made us any more productive at all. The fact that AI makes it easier to draft documents can actually lead to more documents being drafted than necessary, which puts more strain on the system," says Hoppenbrouwers. Zweistra and Hoppenbrouwers also warn of the risk of ending up in a vicious AI circle: "If AI is used for tasks that lawyers are good at, their skills will diminish over time and a situation will arise where the use of AI leads to lawyers relying more on AI."
AI is also seen as an important opportunity to increase access to justice. "We do see that possibility, but we are also cautious. The increase in futile appeals, falsification of evidence, and false reports, which are a major concern for our contacts within the police and the judiciary, is actually placing an extra burden on the system," says Zweistra. "However, we do see opportunities for AI in making legal judgments, for example, fairer and more reliable, which could improve legal certainty and equality."
Zweistra and Hoppenbrouwers emphasize three reasons: "People are afraid of missing the boat. This fear is being increasingly fueled by the industry and providers of legal AI. We see the risk that legal practice will feel compelled to join the AI race without reflecting on the essential question: what problems do I have and can AI solve those problems?" The introduction of AI also leads to new forms of dependency that, according to Zweistra and Hoppenbrouwers, are incompatible with the core values of independence and integrity in legal practice. "In addition, AI is a major driver of the consumption of water, electricity, and scarce raw materials. Moreover, AI is often trained with copyrighted material without obtaining the required permissions. The current use of AI is therefore incompatible with the core value of integrity and the social position of lawyers," says Hoppenbrouwers.
"Faster can be better if, for example, it leads to more cases being dealt with, which could improve access to justice overall. At the same time, speed is not one of the acute problems in legal practice. Careful, fair treatment of cases is a more urgent issue, and the question is whether AI contributes to this," according to Zweistra and Hoppenbrouwers. "Using an LLM to write the reasoning behind a court ruling in order to speed up the process takes away the time the judge needs to reflect on their own judgment. Access to justice is not only about the number of cases that are dealt with. The problem also lies in alienation from the law, in feelings of powerlessness, even when the courtroom has been reached, in not being seen or heard. Time for dialogue and critical reflection is essential in legal practice," says Hoppenbrouwers.
We believe that lawyers should only embrace AI if it offers clear added value. At the same time, we also see that there is a market reality in which it is actually impossible to lag behind. That is immediately the biggest risk: that legal practice will embrace something it does not really need, but that all kinds of new problems and dependencies will be introduced in its slipstream," say Zweistra and Hoppenbrouwers.
Although there may be advantages to be gained from the use of AI in legal practice, Zweistra and Hoppenbrouwers believe that it must be done carefully and, above all, that it must be closely aligned with the problems that lawyers encounter in practice. "Don't be afraid to go against the grain. It is crucial to think carefully about AI, what it does, and how you can use it, but only do so based on a specific question, a problem that precedes the AI solution. Otherwise, you will end up with a solution to a problem that does not exist."
