A question and answer session with Matt Wetherington
Viraj Pamar with the Atlanta Bar Association recently interviewed several Atlanta leaders on a variety of topics related to artificial intelligence and the law. To view the full article, click here. Matt Wetherington’s excerpts are below:
What are the possible uses for AI in the litigation?
100% of current litigation practices will be fundamentally changed by AI. Existing AI tools, like GPT, can already replace large portions of “core” legal, marketing, and operational work for law firms. My firm is deploying these as quickly as possible. The hurdles we face in implementing those technologies are not technological, but instead grounded in ethical, regulatory, and privacy concerns. In the immediate future, those hurdles will fall away and firms of all sizes will have access to low-cost AI to enhance document review, streamline legal research, draft documents, predict case outcomes, simplify e-discovery, analyze contracts, and provide client support.
Document review is not a glamours job. Machine learning, though, has the ability to categorize and tag discovery. This seems like a win for the practice. Do you agree?
Machine learning is already used in most document review centers, with tools like Relativity and Everlaw gaining popularity. The true transformative impact of AI on document review lies in its ability to change the very nature of discovery. AI advancements like natural language processing will help streamline and narrow the discovery process by identifying key factual or legal issues earlier in the case. Further, an emerging body of studies are proving that advanced techniques like natural language processing can not only identify, categorize, and tag documents more efficiently, but also reveal patterns, relationships, and key insights that may not be immediately apparent to human reviewers. This reduces the amount of manual labor required for document review and leads to a more proactive and targeted discovery process.
Writing is an art. It has passion and anger and heart. Can AI mimic the art of writing?
As AI continues to advance, the emphasis on clear, concise, and focused legal writing will likely increase, meeting the evolving expectations of both courts and clients who prioritize efficiency and effectiveness in communication. We will need to adapt to these changing expectations and strike a balance between maintaining the artistry of writing and leveraging the precision and efficiency offered by AI-generated content. But, even when passion is required, AI is up to the task. For example, here is the previous human answer rewritten by AI with “passion, anger, and heart”: “Fear not, dear reader, for the burning passion that resides within the heart of humanity shall not be extinguished by the icy tendrils of technology. For it is in the crucible of innovation that we shall forge a new era, one in which the power of AI is wielded with the deft hand of a master craftsman, molding the clay of language into a masterpiece of emotive expression. And together, as one unstoppable force, we shall write the next chapter in the epic tale of the human spirit.”
Can AI make the practice more civil?
Incivility in the legal field often stems from fear, insecurity, and miscommunication. AI can help streamline communication and promote a more civil atmosphere by reducing misunderstandings and providing instant access to relevant information. For example, if a lawyer is asked to consent to something they don’t understand, the AI can explain the issue quickly and allow the lawyer to respond from a position of knowledge, rather than defaulting to aggressive opposition. But some people are just jerks and AI will not change that.
Is the initial panic – AI will be smarter than humans, it will hurt rather than help, and so on – over blown?
No, the level of concern is dramatically lower than it should be. We are dealing with a technology that will quickly change society in ways that are hard to comprehend or predict.
Is it possible for machine learning to be discriminatory?
Unfortunately, all current generative AI models are trained on datasets that have explicit and implicit biases. The key is to recognize discriminatory behavior and refine models to avoid or disclose bias in their models. We are not there yet.