The impact of artificial intelligence on the legal sector has been the subject of debate for almost two decades, particularly in relation to transactional corporate work, such as M&A. In that time, new technology has provided growing support with searching, reviewing and mining due diligence information and, latterly, transactional document building and workflow generation.
However, with the launch of ChatGPT, there is an undoubted shift change in the opportunities available.
Recently, dramatic (and somewhat reductive) headlines have prophesised the legal profession giving way to technology but, as always, the reality is more nuanced and interesting. What’s more likely is that the impact will rise steadily and ultimately improve the legal services provided by law firms. Some areas of the profession may become less relevant but new roles will be created and clients are likely to see enhanced productivity and a highly skilled profession that can concentrate on the sharp end of its practice.
What is ChatGPT?
In short, it’s an artificial intelligence language model, launched in November 2022. The “G” stands for “generative” which, in simple terms, means that, instead of identifying, sorting and categorising information, ChatGPT learns how to take actions from past data and then generates unique output (such as text or images). In doing so, it can produce natural feeling, original content. It’s not the only model of its kind. Two months after ChatGPT debuted, Microsoft added a similar chatbot to its Bing search engine and, in March, Google launched Bard.
ChatGPT and Law Firms
For M&A lawyers (and the legal sector generally), until recently, A.I. had mainly served to create output using rules-based programming, including the organisation of pre-determined phrases and words within documents or the sorting and categorisation of data. These tools have served as time saving or quality improvement aides.
However, emerging AI technology like ChatGPT promises significant step changes offering AI skills that are transformative in relation to the underlying data, to interpret case law and legislation, assist lawyers in generating legal documentation and potentially reducing the risk (and the cost of risk) of carrying out these sorts of tasks.
As with any nascent technology there are some issues to overcome.
Notably, data protection and client confidentiality, a critical issue on corporate transactions, means that the functionality of the raw tools is limited for the time being. Additionally, the models that are currently available have a tendency to “hallucinate” (i.e., make things up). To address these issues, the legal profession will (and is) developing software that will operate on top of generative AI such as ChatGPT. These will be finely tuned versions that are bespoke for legal work and compliant with GDPR. Startups, such as Harvey, are working towards these goals.
The resulting opportunity is that solicitors will increasingly focus their energy on the development of expertise, the exercise of judgment in complex legal matters, and strategic guidance.
Bearing in mind the milestones that create the need for corporate legal advice, human intuition, empathy, accountability and support are particularly meaningful to clients. Selling a business (often a life’s work), raising funds for a new venture and deploying an equity incentivise scheme to motivate behaviours requires human understanding and AI is unlikely to be able to replicate the services that solicitors can provide. Legal services are not just about organising data. They are about relationships, interpretating and predicting needs and formulating strategies that resonate with stakeholders.
Legal issues with ChatGPT
In addition to how the legal sector will use the technology, there are some important legal issues for users of the generative AI to bear in mind. As mentioned above, there are the data processing issues (which are currently significant). Also, when using ChatGPT, it may not always be straightforward to identify the owner of the copyright. Does it belong to the user or ChatGPT? As the model is programmed to generate content that is unique, the output will be original, but not at the hand of a human.
This is a complicated and unsettled legal issue and will be different depending on jurisdiction. For now, it may be prudent to ensure that the role of generative AI models is acknowledged in the creation of content in order to respect IP rights and to avoid claims of plagiarism.