By: Owen Morris, Director of Enterprise Architecture at Doherty Associates
Imagine having a tool that can generate fresh content at the touch of a button. Enter generative AI, a revolutionary technology that has taken the world by storm. By harnessing machine learning models, generative AI can generate quick content in response to users’ prompts – whether it be text, images or code (and other examples are sure to follow).
The rapid advancement of generative AI has sparked excitement and curiosity across industries. This cutting-edge technology opens up a realm of possibilities, especially as organizations can now combine their own data with public information within the models and effortlessly generate text tailored to their needs. It’s a game-changer for the global workforce, offering a powerful tool to enhance productivity and innovation.
Whilst useful, it’s important to remember that generative AI is a handy tool in your arsenal, not a replacement for your human team and judgement. To harness the true potential of generative AI, it is essential to navigate its capabilities and limitations with finesse, especially within the legal sector.
What makes generative AI valuable to the legal sector?
In the long run, this technology will likely help firms do what they do best, better. By utilizing these solutions, legal firms can generate initial drafts of deals and proposals swiftly. The advantage of this is a rapid turnaround and a competitive edge over rival firms. The key trick will be that, when combined with your data, the model can turn around specific texts trained on your best practices and information.
There are specialists within the field, but we expect general technology vendors to be able to catch up due to partnerships such as Microsoft’s with OpenAI and through their additional computing power and greater access to data through ecosystems such as Microsoft 365 and Google Workspaces. Advancements like Microsoft Copilot enable tools to become more personalized by augmenting them with a firm’s specific data or from other knowledge bases. Responses are then tailored and address each firm’s unique needs. Firms can also use the tool to adapt pre-existing material to work for new scenarios quickly. ChatGPT, for example, can provide instant answers to complex matters, eliminating the need for extensive web searches and saving your team valuable time.
Additionally, rather than spending hours processing large documents and data sets, discovery teams can use the tool to generate a summary of the key points in minutes. In the Covid Inquiry, for example, researchers plan to utilize AI to analyze responses to ensure key trends and insights are accounted for in the reports, which will be submitted as evidence, and speed up the overall process.
However, users must exercise responsibility and possess a comprehensive understanding of the content’s reliability and limitations. The generated text is only as good as the information that feeds the model. Amongst other limitations, such as hallucinations, users still need to interpret the text and decide if it’s valuable.
How can legal firms get started with Generative AI?
To harness the benefits of generative AI effectively, legal teams should begin by experimenting with different tools to understand their functionality and identify the most suitable options. Tracking daily tasks allows teams to identify areas where AI can assist, particularly in unfamiliar legal topics.
Many organizations we work with are investigating how best to leverage AI. A key step is to categorize and document your existing information so that it can be used as source data to augment the AI models. Machine-learning based classification technologies are often used for this (for example, Microsoft Syntex). Just as important is ensuring data that shouldn’t be used with AI is kept private using appropriate security controls.
In parallel, firms may want to consider experimenting with generative AI in three stages to evaluate the benefits for their use cases.
- The first is looking at what a public model tells you. This will give an idea of what is publicly known about the topic and what a non-specialist would think about your prompt.
- The second is to investigate vendors that incorporate specific information from the field – essentially bringing in what the legal sector thinks in general.
- Another stage is to find products that allow your data to be brought into the response, providing a ‘house view’.
- The last is ensuring that the output reflects what you and your firm would generally offer on a similar subject. This is, perhaps, the most important stage, as it requires the user to interpret the generated information and make edits based on experience and knowledge.
- Once a good output is achieved, then it’s important to be able to recreate it in future,
Working in this way gives users texts based on what is publicly known about the topic and your specialist data, filtered through your personal knowledge. In my opinion, this is the most effective way to approach generative AI and should be the foundation for training your team to use models. Getting your team started with quick and effective adoption is crucial to stand out in this increasingly automated landscape.
The challenges of generative AI
Integrating generative AI in the legal field presents both challenges and opportunities. Caution and interpretation are crucial when relying on AI-generated answers, as assessing their accuracy and reliability can be challenging, especially for non-experts. Firms must avoid using it uncritically and without human oversight.
It is vital to view AI as a complementary tool that fosters human-machine collaboration, recognizing that human judgment remains indispensable in the legal profession. Firms need to train their teams to critically assess the output of large language models and compare the output to equivalent human-produced items. Failing to do so and including incorrect data within legal documents risks your firm falling foul of regulations, such as the recent case where a lawyer working in New York is facing legal action following the uncritical use of ChatGPT in legal research. Leaders must ensure they incorporate costs of review and evaluation when looking at business cases for AI.
Furthermore, ensuring data security and compliance poses a significant challenge when utilizing generative AI models. Assessing the risk profile and determining where it can be used safely is essential. Legal teams must also be diligent in safeguarding confidential information and adhering to regulatory requirements. Composing internal policies regarding how to act with trusted information, such as introducing it to the model or incorporating it into your internal datasets, should be a top priority for firms looking to introduce AI solutions.
Knowing the provenance of the data is also crucial to avoid accidental intellectual property infringement. Understanding data provenance and whether corporate data can be used for model training are vital points to cover with vendors.
The next stage
Generative AI has the potential to revolutionize the legal field by enhancing efficiency, accuracy, and competitive advantage. By implementing solutions, legal firms can streamline their processes, generate personalized responses, and stay ahead in an increasingly automated landscape.
However, the responsible adoption and interpretation of generative AI is critical to its success. Legal professionals must develop strategies to overcome the challenges associated with its implementation if they hope to embrace generative AI.
By leveraging the power of generative AI while preserving the essential role of human judgment, the legal profession can embrace the transformative potential of this technology and use it to get ahead.
About the Author
Owen Morris joined Doherty Associates in 2014 and is the Director of Enterprise Architecture. He now ‘owns’ the company’s processes and is responsible for maintaining Doherty Associates ISO 9001 and ISO 27001 certifications. Owen also oversees the Microsoft cloud practice, encompassing Microsoft 365, Azure, and Power Platform, steers the project management function, and leads the cloud applications, data and development team. His wide-ranging expertise also encompasses consultancy and project development and delivery. Prior to joining Doherty Associates, Owen clocked up 11 years’ experience at Logica and Capgemini. He also holds master’s degrees in Chemical Engineering and in Computer Science.