FT: Law firms turn to AI for cost cutting and efficiency

Law firms worldwide are increasingly turning to generative artificial intelligence (AI) to streamline operations and reduce costs, with many either experimenting with the technology or planning to implement it. They aim to use AI to automate routine tasks such as reviewing and drafting contracts and providing preliminary legal opinions.

Despite the excitement surrounding generative AI, which can produce humanlike text, images, and other content in response to simple prompts, firms remain cautious of its limitations.

Australian law firm Gilbert + Tobin has adopted a generative AI question-and-answer feature to offer its lawyers specialized knowledge across various practice areas, including contract guidance. To encourage engagement, the firm last year offered a "bounty" of A$20,000 for innovative ideas on utilizing the technology, receiving over 100 suggestions. Caryn Sandler, partner and chief knowledge and innovation officer, notes a "genuine interest in generative AI" across all levels of the firm for tasks such as drafting, communication, and data synthesis.

Gilbert + Tobin has conducted generative AI "masterclasses" for clients and is developing prototype tools based on the best ideas from its initiative. The firm plans to develop some tools internally and acquire others from external software suppliers, Sandler says.

The risks associated with generative AI, including so-called hallucinations where the technology fabricates information, are well-documented. There have been instances of lawyers relying on flawed AI-generated text for court submissions. To mitigate these risks, Gilbert + Tobin has established "guardrails" to keep client data secure and ensure the AI accesses only vetted information.

MinterEllison, another Australian firm, has developed a generative AI tool that produces a first draft of legal advice in about a minute, with approximately 80% accuracy. This tool, developed with IT company Arinco and based on OpenAI's technology, uses historical legal advice and other resources. Simon Ball, a partner in MinterEllison’s environment and planning team, led the project and highlights the system’s efficiency in summarizing case law and drafting advice. The firm aims for 80% adoption within 12 months.

To ensure accuracy, MinterEllison's tool provides sources for its output, and a senior lawyer reviews the AI-generated drafts before sending them to clients. Ball emphasizes that the tool will only be effective if fed with accurate and relevant information, a responsibility his team diligently maintains.

While MinterEllison uses a traditional time-based pricing model, Ball notes that any changes in pricing would require concrete evidence of AI’s impact on client work.

Some firms are exploring selling their AI technology to corporate legal departments. A&O Shearman is among the first, offering its AI contract negotiation tool, ContractMatrix, to clients. Developed with Microsoft and legal AI start-up Harvey, the tool drafts contracts using existing templates, potentially saving seven hours per contract during negotiations. David Wakeling, head of A&O Shearman’s markets innovation group, underscores the importance of focusing on practical gains, estimating a 20% productivity increase based on feedback.

In China, generative AI adoption faces unique challenges. JunHe partner Pete Zhang says Chinese generative AI software is currently limited to tasks like legal research, analysis, translation, and document summarization, as it lacks the quality needed for drafting legal documents. Data privacy rules, particularly restrictions on cross-border data transfers, further complicate the use of foreign AI tools in Chinese firms.

Despite these challenges and the inherent risks, many law firms believe generative AI will be crucial to their future operations. Sandler at Gilbert + Tobin encapsulates the prevailing sentiment: generative AI offers "enormous opportunity" but also "enormous risks," and firms must keep pace with the rapid developments to avoid falling behind.

[With information sourced from Financial Times]