{"slug": "the-most-important-feature-of-a-legal-ai-might-be-knowing-when-not-to-answer", "title": "The Most Important Feature of a Legal AI Might Be Knowing When Not to Answer", "summary": "An engineer behind Jimmy Knows, a legal AI platform for Massachusetts, argues that the most important feature of legal AI is knowing when not to answer. The system is designed to ask clarifying questions when facts are insufficient, rather than generating polished but potentially misleading responses. This approach prioritizes understanding the user's story over making the user speak like a lawyer.", "body_md": "One of the things that gets overlooked when building AI for legal use is uncertainty.\n\nMost AI products are designed to answer questions.\n\nLegal questions are different.\n\nA person might type:\n\n*“I got hurt at work. Can I sue my employer?”*\n\nThat sounds simple, but it isn't.\n\nWhat happened?\n\nWhere did it happen?\n\nWas the person an employee or contractor?\n\nWas another company involved?\n\nWas the injury caused by equipment?\n\nWhen did it happen?\n\nHas a workers' compensation claim already been filed?\n\nA few missing facts can completely change the legal analysis.\n\nThis is why I think legal AI should not be judged only by how many questions it can answer.\n\nIt should also be judged by how well it handles questions that it cannot safely answer yet.\n\n**\n\n**\n\nLarge language models are very good at producing something that sounds complete.\n\nThat is useful in many applications.\n\nIt can also be a problem in legal applications.\n\nIf the system does not have enough information, generating a polished paragraph can create a false sense of certainty.\n\nThe better approach is sometimes to stop and ask another question.\n\nNot:\n\n“Here is what you should do.”\n\nBut:\n\n“Before I can give you a useful answer, I need to know whether another party was involved.”\n\nThat small difference matters.\n\nFor Jimmy Knows, the idea is to start with what actually happened.\n\nA user can describe their situation in their own words instead of trying to figure out which legal category their problem belongs to.\n\nThe system can then connect the facts to Massachusetts statutes and case law and explain the relevant rule in plain language.\n\nThat creates a different design philosophy:\n\n**Don't make the user speak like a lawyer. Make the system understand the user's story.**\n\nThis also means the AI needs to recognize when the story is incomplete.\n\nFor example:\n\n“I slipped outside a store and got hurt.”\n\nThere isn't enough information yet.\n\nWhere did it happen?\n\nWas there snow or ice?\n\nWas there a visible hazard?\n\nDid the store know about it?\n\nThose questions are not just conversational filler. They can determine which legal rules are relevant.\n\n**\n\n**\n\nThe hard part isn't necessarily generating the answer.\n\nIt is deciding whether an answer should be generated at all.\n\nThat means a legal AI system needs something beyond retrieval and generation.\n\nIt needs a concept of **sufficient facts**.\n\nYou can think of the workflow like this:\n\n```\nUser tells their story\n        ↓\nExtract relevant facts\n        ↓\nIdentify missing facts\n        ↓\nAre the facts sufficient?\n     ↙        ↘\n   No          Yes\n   ↓            ↓\nAsk questions   Retrieve authority\n                ↓\n             Explain rule\n                ↓\n          Explain next step\n```\n\nThis is a much more interesting problem than simply connecting an LLM to a legal database.\n\nThe quality of the system depends heavily on what happens before the final answer is written.\n\n**\n\n**\n\nThere is a tendency to think a good AI assistant should always sound confident.\n\nI think the opposite is often true for legal applications.\n\nA trustworthy system should be comfortable saying:\n\n“I don't have enough information yet.”\n\nThat doesn't make the system less useful.\n\nIt makes the next question more useful.\n\nAnd eventually, that can lead to a much better answer.\n\nThat's one of the ideas behind Jimmy Knows: make Massachusetts legal information easier to access while keeping the distinction between legal information and advice clear. The platform provides information based on Massachusetts authority and connects people with an attorney when they need one.\n\nThe future of legal AI probably isn't about making AI sound more like a lawyer.\n\nIt may be about making AI better at knowing when the facts are not there yet.\n\nAnd honestly, that might be one of the hardest parts to get right.", "url": "https://wpnews.pro/news/the-most-important-feature-of-a-legal-ai-might-be-knowing-when-not-to-answer", "canonical_source": "https://dev.to/jimmyknowslaw/the-most-important-feature-of-a-legal-ai-might-be-knowing-when-not-to-answer-4g0e", "published_at": "2026-08-17 20:20:14+00:00", "updated_at": "2026-08-17 20:44:12.345687+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models", "ai-products", "ai-ethics"], "entities": ["Jimmy Knows", "Massachusetts"], "alternates": {"html": "https://wpnews.pro/news/the-most-important-feature-of-a-legal-ai-might-be-knowing-when-not-to-answer", "markdown": "https://wpnews.pro/news/the-most-important-feature-of-a-legal-ai-might-be-knowing-when-not-to-answer.md", "text": "https://wpnews.pro/news/the-most-important-feature-of-a-legal-ai-might-be-knowing-when-not-to-answer.txt", "jsonld": "https://wpnews.pro/news/the-most-important-feature-of-a-legal-ai-might-be-knowing-when-not-to-answer.jsonld"}}