ИИ для помощников юристов

AI entered paralegal work long before chatbots. In e-discovery, technology-assisted review (TAR), also called predictive coding, uses machine learning trained on attorney decisions to rank documents by likely relevance. A 2012 federal decision, Da Silva Moore v. Publicis Groupe, is widely cited as the first judicial approval of the approach. Today TAR is routine in large cases. Generative AI added new abilities: summarizing depositions and contracts, drafting discovery requests and correspondence, building chronologies, and answering research questions in plain language. Major legal platforms now include such tools, for example Thomson Reuters' CoCounsel and LexisNexis's Lexis+ AI. The central risk is hallucination. In Mata v. Avianca (S.D.N.Y., 2023), lawyers filed a brief citing court decisions that ChatGPT had invented, and the court sanctioned them. Since then, many judges have issued standing orders on AI use in filings. A common misconception is that legal-specific tools, which retrieve real documents before answering, cannot hallucinate. Retrieval reduces errors but does not remove them. A 2024 Stanford study found that commercial legal research tools still produced incorrect or poorly supported answers at meaningful rates. Professional rules still apply. In the US, lawyers must supervise nonlawyer assistants under ABA Model Rule 5.3 and its state equivalents. ABA Formal Opinion 512 (2024) addresses generative AI and covers competence, confidentiality, client communication and fees. Pasting client documents into a consumer chatbot can breach confidentiality. The job is shifting rather than disappearing. Less time goes to first-pass review and blank-page drafting. More goes to quality control, managing review platforms, validating results, verifying citations and writing effective prompts. Paralegals who understand both the law and the tools' failure modes are becoming more valuable to their firms.