L’IA remplacera-t-elle les parajuristes ?

Paralegal work is a bundle of tasks, and AI affects them unevenly. The most exposed tasks involve reading and restructuring large volumes of text: summarizing depositions and medical records, first-pass privilege and relevance review, drafting routine correspondence and filling standard forms. E-discovery has used machine learning for years through technology-assisted review, and generative AI now extends automation to summarizing and drafting. Tasks built on relationships, judgment and coordination are less exposed. These include interviewing clients and witnesses, managing trial logistics, chasing records from providers, noticing what is missing from a file, and knowing how a particular judge or clerk's office works. Professional rules also shape the outcome. ABA Model Rule 5.3 requires lawyers to supervise nonlawyer assistance. ABA Formal Opinion 512 (2024) addressed lawyers' duties when using generative AI, including competence, confidentiality and verifying output. Someone has to check what the AI produced, keep client data inside approved tools and record what was reviewed. That work often falls to paralegals. The skills now in demand include evaluating AI tools and how they fail, writing precise instructions, checking output against source documents, managing matter data and e-discovery platforms, and understanding privacy and security obligations. Roles such as litigation support specialist, e-discovery analyst and legal operations analyst draw on this mix. A common misconception is that the question is all or nothing. The more realistic effect is that each matter may need fewer hours of routine work. That could slow hiring for entry-level roles made up mostly of those tasks, while paralegals who can run and check AI-assisted workflows become more valuable. The size of each effect will vary by practice area and firm size, and published labor projections are estimates, not certainties.