Are we about to see legal systems be based on computer-based decisions, or will humans still part of justice?
Rise of the machines…or not?
Some may argue that the law is black and white, and that a computer can do it better than a human. They may claim that law is based on rules and logic, and that a computer can apply them faster and more accurately than a human. They may point to examples where AI has been used to assist or replace lawyers in various tasks, such as document review, contract drafting, legal research, or even predicting court outcomes.
However, this view ignores the complexity and diversity of legal issues and the human factors that are involved in legal practice. Law is not just a matter of applying rules to facts; it is also a matter of interpreting, arguing, persuading, and balancing competing interests and values. Law is not static; its interpretation evolves and adapts to changing social and economic contexts. Law is not uniform; it varies across different jurisdictions, areas, and domains. Law is not impersonal; it affects and is affected by the emotions, motivations, and relationships of the parties involved.
To illustrate this point, let us consider some landmark cases where legal issues could not be dealt with in a binary way. These cases demonstrate the need for human judgment, creativity, and empathy in legal practice, which cannot be replicated by a computer (today at least).
- In the case of Donoghue v Stevenson (1932), a woman drank a ginger beer that contained a decomposed snail and became ill. She sued the manufacturer of the drink, claiming that he owed her a duty of care to prevent harm. The court had to decide whether such a duty existed, and if so, what its scope and content were. The court held that the manufacturer did owe her a duty of care, and that they breached it by failing to ensure the safety of the product. The case established the modern law of negligence and the concept of the “neighbour principle”.
- Another example of a business case that demonstrated the importance of human interpretation in case law is the case of Carlill v Carbolic Smoke Ball Co. (1893). In this case, a company advertised that their product, a smoke ball, could prevent influenza and other diseases, and that they would pay £100 to anyone who contracted the flu after using their product according to the instructions. They also claimed that they had deposited £1000 in a bank to show their sincerity. A woman, Mrs Carlill, bought and used the smoke ball, but still contracted the flu. She sued the company for the £100, but the company argued that their advertisement was not a serious offer, but a joke. The court had to decide whether the advertisement constituted a valid contract, and if so, whether Mrs Carlill had fulfilled the conditions to claim the reward. The court held that the advertisement was a valid contract, and that Mrs Carlill had fulfilled the conditions. The court reasoned that the advertisement was intended to induce people to buy and use the product, and that the deposit of £1,000 showed that the company was serious about their offer. The court also held that Mrs Carlill did not need to notify the company of her acceptance of the offer, as the performance of the act was sufficient. The court awarded Mrs Carlill the £100.
The Human Touch
These cases show that law is not as simple as true/false, and that a computer can’t do it in the same way a human could. Law requires human skills and values that go beyond rules and logic. Law encompasses different fields and domains, such as conveyancing, contract, business, family, or criminal law, which have distinct features and challenges. Law affects and is affected by the lives and stories of real people, who deserve respect and compassion. A computer may be able to perform some legal tasks, but it cannot replace the role of a lawyer as a counsellor, advocate, and mediator. Law is more than a machine; it is a human endeavour.
In contrast to the current state of legal research, which relies heavily on digital and AI-powered tools, the traditional way of conducting legal research was much more labour-intensive and time-consuming. Before the advent of digitalisation and online databases, law firms had to rely on paper-based records to find relevant clauses and information for their cases. This meant that they had to store and manage huge volumes of documents, such as statutes, case law, contracts, and correspondence, in physical libraries or archives. Finding the right document or clause often required browsing through multiple books or files, using indexes or cross-references, or consulting librarians or experts. The process was prone to error and inefficiency, as documents could be misplaced, mislabelled, outdated, or incomplete.
To conduct legal research for a complex case, law firms would typically assign a team of junior staff, such as paralegals, associates, or interns, to sift through the paper-based records and extract the relevant information. The team would have to reserve a room where they could access and review the documents, often working long hours to meet the deadlines. They would have to sort, organise, annotate, and summarise the documents, and create binders or folders with the case material. They would also have to cross-check the accuracy and validity of the information and update it as new developments occurred. The process was costly and tedious, and often resulted in low-quality or inconsistent outcomes.
With the introduction of digitalisation and searching tools, the legal research process has undergone a radical transformation. Law firms can now access and search millions of documents online; they can use keywords, filters, or natural language queries to find the most relevant and up-to-date information for their cases, such as statutes, case law, precedents, contracts, or scholarly articles. They can also use AI-powered tools to generate legal insights, summaries, or arguments, based on the information they find. The landmark cases I referenced above are examples of the plethora of online data available that can be found using tools such as Copilot. This process is faster and far more efficient, as it can take a matter of minutes rather than months to conduct comprehensive legal research.
The process of searching may appear more accurate and consistent, as it reduces the risk of human error and bias, however humans are able to reason in a way that is not ‘yet’ possible with the existing AI solutions.
Challenges of providing off-the-shelf AI in the legal sector
While generative AI can be an asset for legal research and writing, it also poses some challenges when it comes to providing off-the-shelf solutions, such as Copilot or Gemini, to lawyers. One challenge is the diversity and complexity of the legal domain, which requires high levels of accuracy, specificity, and consistency in the use of language and terminology.
Off-the-shelf AI solutions may not be able to capture the nuances and variations of different legal systems, jurisdictions, fields, and contexts, and may produce generic or inaccurate outputs that do not meet the standards and expectations of legal professionals.
For example, these AI tools may generate snippets that are not compliant with the relevant laws or regulations or may suggest grammatical or stylistic changes that alter the meaning or tone of a legal document. Therefore, off-the-shelf AI solutions may need to be customised or adapted to the specific needs and preferences of each lawyer, firm, or case, which may entail additional costs and efforts.
Another challenge is the difference in the level of experience and expertise among lawyers, and how technology enables them. On the one hand, there are tenured lawyers who have accumulated years of knowledge and skills in the legal domain, and who may be aware of and impacted by the speed of technology, but do not necessarily interact or use it extensively. They probably rely more on their own judgment and intuition and may be reluctant or resistant to adopt new technologies that could disrupt their established workflows or challenge their authority. On the other hand, there are graduate lawyers who are new to the profession and who relish the efficiency and convenience with which they can communicate and collaborate using technology. They can be more open and eager to use AI tools to assist them in various tasks, such as legal research, drafting, editing, or proofreading. However, they may lack the critical thinking and field experience to evaluate the quality and reliability of the AI outputs, and as a result become overly dependent on the technology sometimes possibly without verifying or questioning the results. Therefore, off-the-shelf AI solutions may need to balance the trade-offs between automation and human oversight, and provide appropriate guidance and feedback to the users, depending on their level of experience and expertise.
Grasping the deployment and application of AI tools is critical for businesses. Simply possessing technology or sporadic usage doesn’t equate to leveraging its full potential to transform operational methodologies. While conducting workshops, enhancing security consciousness, and grasping use cases are beneficial, it’s the approach and timing that truly empower optimal enablement results.
Conclusion
There have been several instances where generative AI has created hallucinations or fictitious information, which have been mistakenly used in legal cases. The well know example of a New York lawyer who faced sanctions for citing fake cases generated by ChatGPT in a legal brief. These incidents highlight the need for lawyers to verify the legal insights generated by AI-powered tools, and to apply consideration as to who is both responsible and accountable for information that is provided or co-created with the aid of AI tools.
While generative AI can be a powerful tool for legal research and other applications, it is crucial to verify the accuracy of the information it provides before relying on it in any official capacity. The incidents mentioned serve as a reminder of the ethical and professional responsibilities that come with using AI in legal practice.
Tim Russell
Chief Technologist, Modern Workspace
Office of the CTO
CDW
Pictures by: Sora Shimazaki (Pexels), Tima Miroshnichenko (Pexels), Cottonbro Studios, Diana (Pexels).
