Reed Smith experiments with GPT and searches for Director of Applied AI

Global News

18
Apr
2023

Editor

To further its use of commercial Generative AI-based systems focused on positively impacting the delivery of legal services, Reed Smith has created the new specialized senior-level role of Director of Applied AI. 

This new position will report directly to Reed Smith’s Chief Innovation Officer (CINO), David Cunningham, and focus initially on the development and adoption of Large Language Models and new Generative AI tools to complement its long-standing investments in AI and data science.  

Currently, Reed Smith is running pilot programs of commercial GPT-based systems across its lawyers, Gravity Stack, RED team, global research, KM, and other teams.  The firm is also testing GPT-based systems focused on e-discovery and docketing alerts to challenge them against its existing capabilities. 

“Each of these pilot programs allows us to benefit from the learning model of GPT, but they are private sessions so that confidential data is not shared or used to further teach the public system,” Cunningham said. “Overall, our new Director of Applied AI will serve as a technical and scientific complement to the firm’s in-house experience in applying these tools.  This person will also work with our lawyers and staff to highlight, navigate, and improve AI’s impact on ethics, bias, diversity, and access to justice.”

“This is a very important focal point for us,” Cunningham said.  “So, we are taking the progressive assumption that it is a matter of when, not if, Generative AI becomes a commonplace and essential tool in the business and practice of law.  This role is a complement to our Innovation Lab as it helps to freshly evaluate how our delivery model for legal services should evolve over time.  Collectively, we’ll engage our clients to help guide their journeys while learning how we should design our impact from their perspective.”

Likewise, the firm is committed to developing and teaching its own LLM models, not just relying on commercial products.  “Generative AI is a natural evolution from our use of AI in eDiscovery for over a decade,” Cunningham said.  “We were early adopters for e-discovery, contract intelligence, case analysis, compliance, and even automated drafting of time entries.  As with those cases, we’re quick to absorb the tools to learn their strengths so that we can improve them over time.  However, each took years to become truly useful.”

“With Generative AI, we are encouraging testing and validation so we can best determine where it is appropriately applied, where we need to add context from our knowledge collections or data science program, and where we should avoid its use for now,” said Cunningham.