Law

Disciplinary considerations of generative AI

Note:

Searching law databases is a very nuanced and specific process, which is why learning to do searches in each database is an essential skill. As a law practitioner, it is still very important to do the work of deep-diving into relevant cases, legislation, etc.

• Bias in data: AI models are trained on historical case data, statutes, and legal texts which may contain inherent biases.

• Data privacy: Legal work must adhere to strict confidentiality and data protection standards to safeguard client privacy, so be careful when interacting with AI not to compromise data privacy.

• Deliberation: The legal field values the careful and deliberate crafting of arguments and documents. AI tools should support this process without undermining necessary analytical rigor.

• Authenticity in legal practice: Authenticity in legal advice and argumentation is highly valued. Clients trust lawyers to provide personalised and contextually appropriate advice. AI tools should enhance, not replace, the lawyer-client relationship.

• Interpersonal skills: As AI handles more routine tasks, the ability to communicate effectively, negotiate, and build client relationships becomes even more critical for legal professionals. Lawyers need to demonstrate empathy and understanding to effectively advocate for and support their clients.

Real-world examples of generative AI in the discipline

• Dentons launched "fleetAI," using ChatGPT (based on OpenAI's GPT-4) for legal research, document analysis, and content generation. Dentons has partnered with Microsoft to ensure secure data management, with uploaded data not used for model training and deleted after 30 days. 

• Harvey AI is a generative AI platform designed for legal professionals, and leverages AI models specifically trained on legal material. It is designed to help legal professionals research, draft, analyse, and more.

• Gunderson Dettmer introduced "ChatGD," an internal tool leveraging AI models from various vendors via Microsoft Azure, allowing attorneys to interact with and modify legal documents efficiently. These initiatives aim to enhance client service and operational efficiency through AI-driven solutions.

• Generative AI is prone to hallucination. This is where the models make up facts. Remember that just because it sounds authoritative, it doesn't mean it is correct. Consider the impact if the generated text contains factual inaccuracies. For example, two attorneys in the United States have accused the artificial intelligence chatbot ChatGPT of misleading them into including fabricated legal research in a court filing.

Some prompts to help in the discipline

Note: It is important to understand where the AI tool is drawing data from. There are some tools designed specifically for the law discipline such as casetext from Thomson Reuters.

There are also general purpose prompts in this site that can be helpful. The following list is a starting point for navigating this resource: 

• Suggesting counterarguments 

• Role play 

• Case studies

• Improving written expression 

• Preparing for job interviews 

Use cases in the discipline