ChatGPT has created immense buzz as the potentially disruptive recent technology which will transform the way humans interact with computers. Developed by OpenAI, it is a chatbot with an extraordinary ability to generate human-like responses to prompts. Although it is still in its initial stages, it has fired up debates about its possible application in the legal world especially after almost clearing the bar exam and US MLE exam without any specialized training or reinforcement. This is noteworthy because ChatGPT had an accuracy of over 50% and accomplished 60% in most analyses, with the passing threshold of the USMLE being 60% on average. However, it barely passed the bar exam when tested by the University of Minnesota and failed an attempt tested by Sufolk University.
ChatGPT uses GPT 3.5 to create complex responses in a conversational dialogue. It is distinctive in its application of supervised learning and reinforcement learning to optimize responses. Its utilization of Reinforcement Learning with Human Feedback (RLHF) enables it to process directions and generate human-like text. This is unlike previous chatbots, which lacked the near-sentient nature of ChatGPT. Its claim to fame also involves the use of a diverse set of styles with unprecedented coherence and precision. ChatGPT offers the ability to write conversational answers, assist in research and even compose student essays. This presents a host of opportunities for its application in business and law, particularly owing to its sentient nature.
Can ChatGPT be Used in Law?
The legal industry has for some time been receptive (albeit reluctantly) to the use of AI in law. As for ChatGPT, it can help tackle the daily workload of lawyers. It will reduce the time involved in formulating emails, searching for specific data in contracts and creating ideas for solving a well-defined issue. Technology aids legal practice by increasing productivity and reducing inefficient activities. Intelligent tools such as ChatGPT will improve efficiency by procuring information faster.
Some users have tested the potential of this new chatbot. A lawyer tested the chatbot by drafting a will for a Texas couple with ChatGPT. In the first attempt, it failed to include two witnesses, a requirement of Texas law. When commanded to correct it, ChatGPT rectified the error and provided an updated response. According to the lawyer, this draft was close to what is considered legally acceptable.
However, ChatGPT, like humans is evolving and has also committed grave errors in interpretation. It was asked about the definition of "Anfechtungsklage" (legal challenge) in a German administrative court. While it recognized the Anfechtungsklage as the rescissory action of the administrative court, ChatGPT also asserted that the timeframe for filing a case was imposed by the court. This is false. A court cannot extend the legal deadline for filing a case, which is one month after receiving a negative administrative act. Anyone who had made their decisions based on this information would have made a hazardous error.
While we have all been hearing how AI will revolutionize the legal industry, no one stopped to look what might go wrong here.
The Challenges and Potential Problems of Using AI in the Legal Industry
The use of AI in the legal sector brings about ethical challenges about competence, diligence and oversight. It brings with it a host of new situations that current ethics rules are yet to tackle. Here are some examples:
1. Competence
The American Bar Association recognizes that if technology affects a lawyer’s duty to their clients, it is imperative for them to understand why and how it happens. They have a duty to be competent in the letter and practice of law in addition to maintaining their competence in relevant technologies. This requires them to know the risks as well as the advantages. The rapid adoption and advancement of AI makes it cumbersome for lawyers to constantly update their knowledge in the upgraded technology to fruitfully discharge their duty.
2. Black Box Challenge
When a lawyer sends a question to AI software, it enters what has been dubbed a "black box," where the programme performs its magic and gives the user feedback. But it's challenging for people to comprehend what goes on inside the "black box," or how the specific AI system examines the inputs and outputs to produce the results.
AI is challenging for lawyers. It creates an unprecedented issue of upholding their ethical obligations to competence and diligence. As our reliance on technology grows and injustices occur, these issues will persist.
The lack of transparency by tech corporations regarding the inner workings of the AI algorithms is the other problem with the AI black box. Instead of developing their own AI software, the majority of law firms currently rely on third-party vendors. However, the operations of AI companies that develop "black box" AI technologies are frequently opaque. AI businesses may have valid worries about rival businesses stealing their trade secrets or hackers attacking their software. But before deciding to invest in and rely o`n AI software, law companies must take these risk concerns into account and manage them.
3. Bias in AI
AI software is overseen by humans, and we all have biases. Furthermore, prejudice can still exist in AI systems despite their unique approach to problem-solving. The outcome will also be skewed if the data we give AI software is prejudiced or if the system's processes for processing data are faulty.
AI technology, for instance, has demonstrated bias in hiring. Due to its preference for men over women, Amazon discontinued a method it developed to evaluate job applicants in 2018. Researchers have found evidence of racial bias in some of the algorithms that judges may employ to set sentences for defendants of color.
Humans will need to undertake evaluations to ensure algorithmic accountability because there is a high possibility for injustice for underrepresented and weaker groups. Attorneys that are tech savvy may be the ideal candidates for this position. To shield customers from the potential repercussions of relying on AI technology, lawyers may also need to lobby for legislative changes. Again, it is obvious that while AI may lessen opportunity in some areas of law, it will raise opportunity in others.
4. Machines Cannot be Fully Trusted
Small errors in the hardware or software can result in a massive disaster. Although AI might theoretically be error-free, a device or piece of software's error-freeness is not guaranteed. The potential harm in the instance of legal firms could wreak havoc. In this case, the expense of installation is the responsibility of the business. The use of AI tools is not subject to any laws or restrictions. AI is also incapable of listening, empathy, advocacy, or political understanding.
5. Vulnerable to Data Breach
Privacy and cybersecurity are a reasonable concern associated with the implementation of AI. A malpractice insurance recently performed research that found that 22% of law firms were impacted by hackers. Contrary to popular belief, the victims included prominent corporate personalities. However, even smaller companies might fall victim. The American Bar Association reported that this number was 35% among law firms with 10 to 49 practitioners, indicating that more than a third of small law firms had experienced hacking.
Data provided to ChatGPT could be breached thus raising data privacy concerns. This can create grave implications for privacy in legal proceedings.
The following ways illustrate how ChatGPT could be a useful tool for law firms and their clients:
1. Highly optimized Chatbot:
ChatGPT can be developed into a highly optimized chatbot with factual answers provided in detail to the most common and repetitive questions. This would provide speedy and accurate answers in a conversational format.
2. Legal Research:
ChatGPT can speed up the process of procuring information for research purposes. It can further provide:
· Legal clauses
· Legal precedents
· Detailed, factual responses based on a country’s or state’s law
It will respond to follow-up questions, without diving into too much detail, with useful answers.
3. Drafting a Variety of Legal Documents:
ChatGPT can generate several types of legal documents-wills, contracts and Non-Disclosure Agreements (NDAs). This will save the law firm time and thus improve efficiency. Delegating this function to an AI chatbot also ensures reduced errors and a more streamlined process.
4. Analyzing and Reviewing Documents:
ChatGPT can also prove useful for handling document review and analysis. Using this technology, law firms will be able to analyze large heaps of documents in a matter of minutes. This can hasten the process of problem solving and help them devise an appropriate course of action.
For lawyers, ChatGPT offers automation, simple access to information, and precise case prediction. More importantly, AI allows legal teams to offer their clients higher-quality services because it saves a law firm time and money.
ChatGPT increases the accessibility of legal aid for those with little financial means. The Legal Service Corporation reported in 2022 that 92% of Americans with low incomes don't get enough assistance for their serious legal issues. This can be altered using ChatGPT. Users can receive insights and guidance for urgent legal concerns and know what to do thanks to this AI solution. However, it is important to note that ChatGPT can only take on the basic tasks. ChatGPT cannot replace the expertise of a lawyer.
Conclusion
Changes and evolution in legal practice can be unsettling and cause worry. However, in this case, technology could enhance the legal profession's career and expand the number of individuals who have access to justice. ChatGPT gives businesses the tools they need to stay competitive in the legal market. Additionally, it enables less expensive legal resolution for individuals with limited financial resources.
However, ChatGPT is not a wholesome and up to date database of information. Sometimes ChatGPT provides answers that are erroneous or illogical. Additionally, ChatGPT' has knowledge cutoff in 2021, ie., its information repository is restricted to data published till 2021, means that it will no longer be able to deliver correct information on topics and changes , However, there are currently no terms of use, and liability concerns which will undoubtedly need to be resolved before ChatGPT can be effectively used. For tomorrow, human attorneys will continue to present arguments, render judgements, and draught legal papers. It cannot replace a lawyer’s talent, but only try to enhance it.
References
Comments