Free CLE on Generative A.I. Mechanics and Practice

Session Topics:

• Math and Mechanics of Generative AI: Explains the foundation of Large Language Models (LLMs) as next-word predictors, using machine learning, GPT, and reinforcement learning to create multi-dimensional models for text prediction.
• Generative Pre-Trained Transformers (GPT): Describes how GPT reduces large text data into a single vector using tokenization, embedding, and attention mechanisms.
• Reinforcement Learning: Highlights the process of aligning models through human evaluation, automation, and fine-tuning.
• Software Integration: Discusses how refined models interact with broader software systems, including input systems, pre-prompting, and monitoring outputs.
• Using LLMs: Emphasizes the importance of understanding LLM limitations, such as lack of context, common sense, and training data constraints, while highlighting the role of prompt engineering.
• Legal Use Cases: Outlines applications of generative AI in law, including drafting, editing, brainstorming, organizing, search, translation, and cite checking.
• Legal Pitfalls: Warns of unauthorized practice of law by non-lawyers using AI and highlights the importance of adhering to professional conduct rules, competence, diligence, and candor.
• Confidentiality Concerns: Discusses risks of sharing sensitive information with third-party AI models and provides strategies to prevent confidentiality breaches, such as opting out of data ingestion and using locally run models.
• Bias and Prejudice: Addresses bias in training data and prompts, emphasizing the need for competence in using AI tools to avoid missed insights and incorrect inferences.
• Generative AI Checklist: Provides key recommendations, including understanding the technology, educating firm members, using AI as a secondary source, preserving confidentiality, and being mindful of biases in prompts