Generative AI and the Legal Industry: Q&A with Expert Owen Morris, Operations Director at Doherty Associates. Owen has been on the forefront of how generative AI can be used in law firms, transforming the landscape. With a wealth of knowledge and experience within the legal sector, Owen also discusses some of the challenges and opportunities this technology presents. This interview should hopefully serve as a guide for legal professionals looking to embrace this technology and set their firm apart in a competitive industry.
How are law firms using generative AI?
“Law firms are utilising generative AI in different ways, with varying levels of adoption and progress. Almost all UK law firms will have an AI project or plan being worked on. Harvey AI, for example, is one of the largest in development, with Allen & Overy using it since 2022. Clients frequently seek guidance on incorporating AI into their operations, while others are already further along in their AI journey. Although the tools are not yet highly specialised for individual firms, they are tailored to meet the needs of the legal industry.
“Generative AI provides a starting point by generating initial text that humans can refine, although human intervention is still crucial to ensure accuracy due to the potential for hallucinations. Law firms can leverage generative AI to create proposals, repurpose well-written content, conduct initial searches, and generate boilerplate text. The next phase will be augmenting models with data from various sources and enabling a more personalised touch.”
What are the examples of generative AI?
“There are many different types of generative AI that produce different types of output, for example, text generation (like ChatGPT and other large language models) or image generation that approach the level a human can produce. It’s a quickly changing field, and many vendors are incorporating this technology into their general products while other specialists are trying to bring tuned models to areas like legal firms.
“The value of generative AI depends on how it is deployed. Some businesses focus on prompt engineering, augmenting models with external data, or even training their own models to tailor the technology to their specific needs. This customisation allows businesses to leverage the unique benefits of generative AI in a more targeted and specialised manner.
What can generative AI be used for?
“Generative AI has many uses and applications, and new applications are being developed all the time! In the legal space, it’s principally used for document generation and summarisation applications. It provides a solid starting point by generating initial drafts based on user commands, which can save time and effort. Generative AI is also useful for obtaining general information within a specific domain, as it can deliver tailored responses based on refined research.
“Moreover, it integrates well with other machine learning techniques, combining different approaches for enhanced performance. Generative AI excels in document search and categorisation tasks, facilitating efficient information retrieval. Additionally, by offering generative AI as an API, it becomes a gateway technology that can foster increased adoption of tech solutions within the legal industry.
“It’s also becoming cheaper for organisations to train novel models, and even feasible for companies to develop their own, so it’s a high-speed field. As the barriers to entry for technology continue to lower, generative AI holds the potential to unlock new possibilities and efficiencies in various domains.”
What is the best AI application for lawyers?
“There are several applications that can be useful for lawyers. Tools like eDiscovery, which can summarise contracts, transcripts or legal briefs, are particularly helpful for document generation and highlighting key points & recurring themes. Matter management is another key area with applications capable of summarising discussions from face-to-face meetings and conference calls.
“There are also other wider applications outside generative AI that are important, including document classification, topic identification, document search, transcription and OCR, that use machine learning techniques to boost the accuracy of these important functions for legal firms.”
What are the disadvantages of AI in law firms?
“While AI offers numerous benefits, there are certain disadvantages that law firms should be aware of. One risk is the uncritical use of AI, relying on it without thorough evaluation or critical analysis. As generative AI is still in its infancy as a technology, not every piece of content generated is accurate. It’s essential that human oversight is maintained. Recent examples of the misuse of generative AI within witness statement creation offer a cautionary tale of the importance of mindful use of this tool.
“Another concern is the potential for AI to replace human professionals, which can affect job roles and expertise. However, generative AI is being deployed to augment and support human positions and not supplant them. Whilst generative AI tools like ChatGPT offer internet-based insights, human involvement remains essential for accuracy and value addition. It’s important to remember to use your team to contribute their legal-specific data to enhance the models’ applicability. A business risk worth considering is whether the gains in generating content are greater than the effort taken to check the output!
“Compliance professionals may be particularly concerned about technology that risks introducing unvetted information into a system. Law firms must establish clear policies on using and managing AI to mitigate risks effectively. The lack of clarity regarding the sources of information and data used by AI systems further raises the possibility of copyright infringement, especially with image and code generators. From a security standpoint, assessing the data usage and potential risks associated with AI technologies is crucial, as some AI vendors will incorporate any text that is input into their data training sets. Preparing and training your team to manage these risks is essential for all firms looking to leverage generative AI.”
How do I create a generative AI?
“One option is to utilise a vendor-supplied program that already integrates generative AI, either tailored to a specific industry or as a general-purpose solution. Examples include Adobe Smart Fill or AI-generated backgrounds in PowerPoint.
“Another approach is to leverage existing models like ChatGPT or similar offerings from different vendors through APIs, seamlessly integrating them into your existing workflows.
“Augmenting publicly available models with your own data can provide a unique value proposition since it adds organisational knowledge not available to outsiders. This is currently hard but not impossible to achieve, but vendors like Microsoft, with their Copilot tool, will be making this capability more widely accessible. It’s also becoming cost-effective to fine-tune open models with company data, and sophisticated organisations are looking at this.”
What is the difference between generative AI and LLM?
“Generative AI is a broad term encompassing using AI to create new content. On the other hand, LLMs are a specific form of generative AI specifically focused on generating text from text. The model is trained on a very large input data set plus a very large neural network, through which the model is able to identify entities and concepts in such a way that the output text is very close to what a human would write.
“This is distinct from other applications of generative AI, such as image generators, where you input text and receive an image as the output. LLMs excel at generating text-based responses and have become valuable tools for various applications, including natural language processing, text generation, and conversational interfaces.”
Can generative AI write code?
“Generative AI has made significant advancements in code generation. GitHub Co-pilot is a prominent player, offering a powerful code generation tool. Recently, Salesforce also introduced a similar solution. These AI-based tools have the capability to generate code by utilising existing public code repositories.
“However, it is important to note that some concerns have arisen regarding the verbatim generation of code that may be licensed or copyrighted. As generative AI continues to evolve, navigating legal and ethical considerations associated with code generation is crucial to ensure compliance with licensing and intellectual property rights.”
How is AI affecting law firms?
“Whilst AI may not currently have a significant impact on law firms, many recognise that AI is here to stay and is a valuable tool if properly harnessed. Failing to embrace and incorporate AI technology may result in firms falling behind competitors. So, as a result, lots of firms are actively exploring how to leverage AI and the associated risks and implications.”
Why is AI good for law firms?
“AI provides a head start for projects, allowing law firms to make significant progress at a faster pace. This acceleration can be particularly advantageous in complex legal projects and research. Additionally, AI assists in finding valuable resources and insights within the firm’s practices, enabling lawyers to leverage machine learning, AI-powered business intelligence, and other related technologies.
“It’s important to recognise that generative AI is just one of the many tools available in the AI landscape. Law firms should continue exploring and adopting various AI applications to enhance their operations and stay competitive in the legal industry.”
Can AI be problematic in the legal sector?
“One of the critical concerns is the risk associated with the provenance of data and the potential lack of human oversight. AI systems rely heavily on data inputs, and if the source and quality of the data are not carefully monitored, it can lead to biased or inaccurate results.
“Internal access to data is something firms need to consider. For example, Verein law firms have challenges accessing internal information, as it is restricted to the constituted parts of the organisation. This makes for very complex permission structures or no access at all. Furthermore, many legal professionals work on a consultancy basis, meaning that access to data and systems needs to be tightly controlled with clear boundaries.
“Additionally, the lack of human oversight in AI processes can raise ethical and legal concerns, especially regarding decision-making or providing legal advice. The legal sector must implement robust governance frameworks and ensure human involvement and accountability in AI applications to mitigate these potential problems and uphold the integrity and reliability of legal services.”
What’s next for generative AI?
“The future of generative AI is an exciting prospect. One significant development in the market is the emergence of vendors that enable users to train models using their own data. This opens possibilities for customised and domain-specific applications. Another area of advancement involves integrating different models, enabling more sophisticated and context-aware responses.
“Besides leveraging general data, using proprietary data to generate responses will likely gain prominence. As generative AI continues to evolve, the importance of staff training will grow. Lawyers, who possess a strong command of language and its nuances, are well-positioned to harness the potential of generative AI effectively. By embracing these advancements, the legal sector can unlock new productivity, efficiency, and innovation levels.”
About our expert
Owen Morris is the Operations Director at Doherty Associates. Owen has been on the forefront of how generative AI can be used in law firms, transforming the landscape. With a wealth of knowledge and experience within the legal sector, Owen’s insights above illustrate some of the challenges and opportunities this technology presents.
This interview should hopefully serve as a guide for legal professionals looking to embrace this technology and set their firm apart in a competitive industry.
We thank Owen and his team very much for these considered replies and would recommend anyone interested in this subject follows Owen on LinkedIn here.
And you can also read WardblawG’s guide to AI software for law firms here.