How AI is set to transform the legal sector [Q&A]
The use of generative AI is becoming common across many industries, but while it undoubtedly offers benefits it can lead to problems too.
Legal firms in particular can fall foul of poor results, one firm was fined $5,000 after a court found that one of its lawyers had used ChatGPT to write a court brief which included false citations.
On the other hand it can be useful in streamlinging workflows such as drawing up contracts. We spoke to Josh Hogan, senior product manager (AI) at NetDocuments, to find out how legal firms can employ AI in a secure way, protecting client confidentiality and ensuring security and compliance.
BN: What benefits can AI bring to the legal field?
JH: Generative AI using the new breed of large language models (LLMs) promises to be a real game changer for the legal sector. Ultimately, it will make lawyers' lives much easier, saving them copious amounts of time. For example, it will make drafting documents, such as contracts, engagement letters and retainer agreements, way more efficient. Documents such as depositions and complaints could also be instantly summarized. And instead of just extracting data points like courts, addresses and jurisdictions, legal professionals could turn text into actionable insights that help clients to understand and track their legal obligations.
Ultimately, generative AI has the potential to transform the legal industry by not only automating mundane tasks but also by providing valuable insights and intelligence. This will bring many benefits, leading to more informed decision-making, increased efficiency, and overall improved client services.
However, despite the benefits the technology is set to bring, it is important for law firms and legal teams to understand the implications of using generative AI in legal workflows. Decisions around implementation of AI for legal organizations must be taken after due consideration and the technology itself must be implemented responsibly so that security, client confidentiality, intellectual property, and governance are maintained.
BN: What are the implications of using generative AI in the legal sector?
JH: The potential of generative AI is undeniable, and the legal industry is arguably one of the best positioned to benefit. The challenge for law firms and other legal organizations is that, while the number of actual or future offerings announced grows on a seemingly daily basis, few seem to offer the degree of control and flexibility required to support the needs of legal professionals at scale.
Many of the AI solutions that are currently available in the market are essentially thin application layers built on top of publicly available chatbots or retail AI models. They lack the ability to intelligently and consistently design prompts and incorporate crucial context, making it difficult to produce accurate and valuable results. Many require the user to become an expert prompt engineer to get useful and accurate outputs. Others conceal the critical details of prompt design and application parameters in a 'black box' -- meaning law firms and legal departments cannot easily modify application behavior and outputs to meet their specific needs.
Without the appropriate balance of control and flexibility, legal professionals will be unable to reap the full benefits of generative AI. In fact, productivity may be hampered as time will have to be spent on reworking outputs from AI-powered applications. At the same time, users of applications that do not include the specific security, compliance and governance features required by legal organizations may face challenges responsibly integrating the technology into their mission-critical workflows.
In short, law firms and legal departments must be diligent in developing ways to responsibly utilize generative AI, while reducing risk to their organizations, clients, and security.
BN: How can you make the results that generative AI produces more accurate?
JH: To provide more accurate results, precedent must be embedded into the AI prompt. However, legal firms and departments need to be aware of the privacy and security considerations that come with exposing this proprietary material and confidential data to an AI model. For example, firms and legal teams should be asking themselves whether it is safe to share the context of their sensitive documents and data -- or their clients' data -- to take advantage of these capabilities. And whether the data shared will be retained, and for how long.
The most valuable generative AI use cases arise from leveraging a firm's repository of documents and associated data -- whether to search that corpus, classify documents, analyse and extract data from its contents, or generate new drafts based on prior precedent.
BN: So how can law firms use generative AI responsibly?
JH: While leveraging chatbots like ChatGPT to generate content is easy, integrating advanced LLM AI content generation into legal workflows at scale and in a secure and compliant manner is much more difficult. Any legal organization must ensure the following are true:
- Any information sent to, or received from, an LLM is fully secured (e.g., encrypted) and subject to granular access controls that define who can view the information, for what purpose, and for which clients or matters.
- None of the information sent to, or received from, the AI model is used to train the base model itself.
- None of the inputs or outputs are subject to automated or human oversight by any third party, including the AI vendor and any provider(s) of AI-enabled software.
Law firms and legal teams should look to integrate generative AI with their secure document management system (DMS) that already houses the confidential content firms and departments maintain. Doing this will provide the requisite encryption, access controls, data loss prevention features, security analytics, and ethical walls that are needed to remain secure. This will give legal organizations the ability to work with the entire corpus, conform it to their security and compliance regimes, and integrate it seamlessly with the workflows and productivity tools that lawyers use every day.
BN: And lastly, what overall advice would you give to law firms looking to use generative AI?
JH: This is an exciting time for the legal sector. It's understandable that legal organizations will be wanting to move fast and use this technology to remain competitive. However, it must be done responsibly. Security and privacy simply cannot slip through the cracks. To reap the benefits of generative AI, legal organizations should take a thoughtful, step-by-step approach that has a true commitment to security.