Sign In
    Wisconsin Lawyer
    February 07, 2023

    Technology
    ChatGPT Artificial Intelligence: Will It Replace Lawyers and Legal Staff?

    ChatGPT technology has the potential to be a milestone for technology advancements in the legal profession, as seen with the internet, online legal research platforms, and cloud-based case management systems. It is a tool to help lawyers more efficiently deliver legal services.

    Christopher Cody Shattuck

    robot using computer

    Imagine you want to write a letter to a potential client declining representation due to a conflict with a former client, but you are having trouble finding the right words. Now, imagine a scenario where you can simply type your thoughts into a search bar, like Google, and have the search engine help you generate ideas and language for the letter to your client. Chat Generative Pre-trained Transformer (ChatGPT), using GPT-3.5 language technology, is the next wave of artificial intelligence software that can do just that, and much more.

    How Does it Work?

    Launched and freely made available to the public at the end of November 2022, the chatbot defines itself as the following: a language processing artificial intelligence model developed by OpenAI, capable of generating human-like text, and can be used for a variety of natural language processing tasks, such as language translation, summarization, and question answering.

    At least, that is the summarized version of the general definition you receive when you visit https://chat.openai.com/, log in, and then type “what is ChatGPT?” into the search bar interface. If you want to know how artificial intelligence powers the process, ask ChatGPT and it will respond with something like the following:

    • ChatGPT works by first “pre-training” a large Transformer model on a massive dataset of text. This pre-training step helps the model learn the patterns and structures of language, which it can then use to generate text when it is fine-tuned on a specific task.

    • Once the model has been pre-trained, it can be fine-tuned for a variety of language-generation tasks. For example, a user could use ChatGPT to generate a document by providing the model with relevant information and asking it to generate text based on a specific prompt.

    • One of the key benefits of ChatGPT is its ability to generate highly coherent and contextually appropriate text. It can understand the context of the text it is generating and use that information to make decisions about which words and phrases to use. This makes it an extremely powerful tool for professionals who need to generate complex documents that require a high level of language accuracy and precision.

    You might be thinking that a simple Google search can answer questions and that Google hasn’t replaced the work of lawyers and their staff. However, ChatGPT can do much more than merely answer simple questions and is more powerful than Google. For example, ChatGPT can generate client communications and legal arguments for use in brief writing and summarize legal content for marketing purposes (and it even helped edit the first draft of this article).

    ChatGPT Examples

    To demonstrate the capabilities of ChatGPT, I have included three examples of inputs given to ChatGPT and the corresponding responses below.

    Christopher C. Shattuck Christopher C. Shattuck, Univ. of La Verne College of Law 2009, M.B.A. U.W.-Oshkosh 2015, is manager of Practice411, the State Bar’s law practice assistance program. If you have questions about the business aspects of your practice, call (800) 957-4670.

    Example 1 – Input: Write a letter to a prospective client explaining that my firm will be unable to represent them due to a conflict with a former client and cite Wisconsin Supreme Court Rules.

    Response: Dear [Prospective Client],

    Thank you for considering our firm to represent you in your legal matter. After carefully reviewing your case, we have determined that we are unable to take on your representation due to a conflict of interest with one of our former clients.

    As attorneys, we are bound by ethical obligations to avoid conflicts of interest and to zealously represent our clients. One of these obligations is outlined in Wisconsin Supreme Court Rule 20:1.7, which prohibits us from representing a client if it would create a conflict of interest. In this case, our representation of you would create such a conflict, and we are unable to take on your representation as a result.

    We understand that this may be disappointing news and we apologize for any inconvenience this may cause. We recommend that you seek representation from another law firm.

    Thank you for considering our firm, and please do not hesitate to contact us if you have any questions or need further assistance.

    Sincerely, [Your Name]

    Example 2 – Input: Write a 300-word legal brief in support of a defendant’s motion to extend the discovery deadline response timeframe in Wisconsin. Plaintiff will not extend the deadline and defendant is ill. Include citations to Wisconsin statutes.

    Response: I am writing to request an extension of the deadline to file responses to the plaintiff’s discovery requests. The current deadline is [date], but the defendant is unable to respond timely due to illness. The plaintiff has not agreed to extend the deadline and has not taken any steps to accommodate the defendant’s situation.

    Under Wisconsin Statute § 804.12(2), the court has the discretion to extend the time for discovery upon a showing of good cause. In this case, the defendant has a good cause for requesting an extension because they are unable to respond due to illness.

    In addition, Wisconsin Statute § 804.01(3) provides that the court shall consider the need for a party to have a fair opportunity to respond to discovery requests in determining whether to grant an extension. In this case, the defendant will not have a fair opportunity to respond to the plaintiff’s discovery requests unless the deadline is extended.

    Granting an extension in this case would not prejudice the plaintiff and would ensure that the defendant has a fair opportunity to respond to the plaintiff’s discovery requests. This is in line with the principles of fairness and justice that guide the discovery process in Wisconsin.

    In light of the defendant’s illness and the plaintiff’s refusal to extend the deadline, granting an extension of the discovery deadline is the only fair and just solution. It would ensure that the defendant has a fair opportunity to respond to the plaintiff’s discovery requests and allow the parties to fully and fairly present their case.

    For these reasons, the defendant respectfully requests that the court grant an extension of the deadline to file responses to the plaintiff’s discovery requests.

    Thank you for your consideration.

    Example 3 – Input: I am an attorney practicing civil rights law in Wisconsin and I want to create content on my website. Summarize Brown v. Board of Education into a blog post that readers without a legal background can understand. Please also explain how my experience representing plaintiffs in state and federal courts for over thirty years helps make me a great attorney for their case.

    Response:Brown v. Board of Education was a landmark 1954 Supreme Court case that declared segregation in public schools to be unconstitutional. Prior to this decision, many schools in the United States were separated by race, with black students attending separate, often inferior, schools from white students. The plaintiffs in the case were a group of African American parents who argued that segregation in education denied their children equal opportunities and violated the Equal Protection Clause of the 14th Amendment to the Constitution.

    The Supreme Court ruled in favor of the plaintiffs, finding that segregation in education was inherently unequal and violated the Constitution. This decision helped to pave the way for further civil rights efforts, including the Civil Rights Act of 1964 and the Voting Rights Act of 1965.

    As an attorney with over thirty years of experience representing plaintiffs in civil rights cases in both state and federal courts, I have a deep understanding of the laws and issues surrounding discrimination and inequality. My experience and knowledge make me well-equipped to advocate for my clients and fight for their rights. Whether you are facing discrimination in the workplace, at school, or in your community, I am committed to providing you with the legal representation you deserve and fighting for justice on your behalf.

    Limitations

    ChatGPT adjusts its responses based on the inputs it receives and can generate different responses when asked. However, there are exceptions for certain types of questions, such as general queries like “Why is the sky blue?”, which will typically have the same response, unless you also ask ChatGPT to explain the response to a five year old or in a speech to scientists.

    ChatGPT can also learn and adapt within an existing conversation. For example, a user can create a fact pattern for ChatGPT to detail the requirements for creating a will in Wisconsin and then ask ChatGPT to create the will. If feedback or revisions are needed, ChatGPT will consider the input information provided by the user in the message history and its training before providing a response.

    Despite its capabilities, ChatGPT does have limitations that I discovered while writing this article and experimenting with the artificial intelligence software. In one scenario, I asked ChatGPT to generate a defendant’s motion to dismiss a foreclosure suit in Wisconsin; the plaintiff had improperly served the defendant by publishing notice in a county that was not near where the defendant worked or lived. ChatGPT created a short, concise, and accurate legal argument but cited fabricated case law in Wisconsin. Specifically, it cited a nonexistent court of appeals case in Wisconsin and created quotes from the nonexistent case to support the legal argument. ChatGPT is a language-processing artificial intelligence model and it will attempt to generate a response based on its training, even if the response is not accurate. While it should not be used to generate legal arguments, if it is supplied with a legal inquiry, it will attempt to respond the best it can, even if that means creating fabricated case law that does not exist to fill in the gaps for its response.

    In another example, I asked ChatGPT to generate a client engagement letter that included the terms of the alternative-protection-for-advanced-fees provision outlined in Wisconsin Supreme Court Rule 20:1.5(g). In this scenario, the client agreed that the law firm could deposit unearned advanced fees paid by the client into the law firm’s business account, rather than a trust account. However, the response generated by ChatGPT failed to understand the rule and, as a result, did not include in the engagement letter the necessary requirements for the alternative protection for advanced fees. The letter, therefore, was not accurate and did not properly reflect the nuances of the rule.

    ChatGPT does transparently explain its limitations when queried. These include:

    • Limited understanding of the context and background of provided information,

    • Inability to browse the internet to verify information’s accuracy, and

    • Limited understanding of the real world because its understanding is based on the text available during its training and its knowledge cutoff is 2021.

    ChatGPT also explains that it might make errors, particularly when interpreting idiomatic expressions and metaphor, and when generating content that requires more advanced knowledge than what is available in its programming; and that it might have biases because its training data is sourced from the internet, which can perpetuate certain biases in the real world.

    ChatGPT further explains that it should not be used as a substitute for professional legal advice because the laws and regulations in different jurisdictions are complex, and the facts and circumstances of each situation can affect the application of the law. Furthermore, ChatGPT emphasizes that to provide accurate legal advice, a lawyer must have a thorough understanding of the relevant laws, regulations, and case law and the facts and circumstances of a particular case. ChatGPT recommends consulting with a qualified lawyer who is licensed to practice in the appropriate jurisdiction because those attorneys will have the knowledge, skills, and tools to provide the best legal advice possible.

    While there are applications that can help predict whether text was generated by ChatGPT technology,1 the applications only provide predictions and are unable to fully guarantee that the text was not written by a human.

    Despite its limitations and relatively recent release, ChatGPT is already making strides in the legal industry, generating discussions, articles, and investments.

    Real-World Applications

    Can GPT technology be used to pass the multiple-choice section of the Multistate Bar Examination? Not yet, but researchers expect that the newer version of GPT-4, expected to be released in 2023, will be able to pass the exam.2 What about writing a law article? Yes, Reuters reported that Suffolk University Law School Dean Andrew Perlman used ChatGPT to coauthor a 14-page law article in one hour.3

    Countless other experiments with the technology have been published in articles and blog posts, on topics such as using ChatGPT to write computer code, essays, legal briefs, and syllabi, draft contracts, and conduct legal research.4

    If the current use examples are not enough to catch your attention, even trillion-dollar companies such as Microsoft and Google are considering the effects of ChatGPT on their future revenue streams and ways to incorporate it to add value to their existing platforms. Legal research provider LexisNexis has also conducted experiments with ChatGPT and found it helpful for answering questions but not reliable enough to be used for legal research.5

    ChatGPT is not ready for prime-time use by attorneys in their everyday legal research and brief writing. But for self-represented litigants, ChatGPT is a more interactive tool than online self-help guides and internet searches. Self-represented litigants could upload a summons and complaint, motions, discovery requests, and notices into ChatGPT and ask for explanations or for responses to be drafted.6

    Predictions for Law-related Uses

    It is likely that ChatGPT will be embedded into the technology attorneys use every day in their practices. I think the future technology will allow lawyers to upload briefs, complaints, or pleadings into legal research platforms and request that the platforms generate responsive documents that cite applicable facts, statutes, and case law. The technology also will enable lawyers to type in facts, upload documents, and provide jurisdictional information for the legal research platforms to generate initial briefs, complaints, and pleadings. ChatGPT likely will also be embedded into future versions of word processing, email, and spreadsheet programs that will help users quickly create content or double-check the accuracy of their work.

    The ability to automate the work product of attorneys and other legal professionals will undoubtedly result in the reduction of the total hours needed to complete tasks. There probably will be a wave of new ethics and court rules, statutes, and legal decisions meant to address the use of ChatGPT technology. (For a discussion of the effect of the technology on the current professional rules of conduct, see the accompanying sidebar.) These advancements should not be met with resistance or fear; they will present an opportunity for attorneys to have more time to focus on billable hours and the higher-level skills needed to best represent clients.

    Conclusion

    ChatGPT technology has the potential to be a milestone for technology advancements in the legal profession, as seen with the internet, online legal research platforms, and cloud-based case management systems. To truly understand and appreciate the technology, lawyers should try it for themselves. As you do, consider the effects the technology will have on your practice and the corresponding ethical implications. Try to keep an open mind about this technology: Computers will never replace human lawyers. Instead, computers will continue to have the same effect on the legal profession: automating processes to help lawyers more efficiently deliver legal services.

    Ethical Obligations When Using ChatGPT

    Lawyers using ChatGPT must carefully manage nonlawyer assistance, protect confidentiality, provide competent representation, exercise independent professional judgment, verify the accuracy and authenticity of text and citations generated by the software, and perform other duties owed to clients and third parties.

    Aviva KaiserBy Aviva Meridian Kaiser

    When a lawyer uses nonlawyer assistance, such as artificial intelligence-powered legal research or a language processing artificial intelligence model, the lawyer is bound by SCR 20:5.3, the lawyer’s responsibilities regarding nonlawyer assistance. ABA Comment [3] following SCR 20:5.3 cautions: “When using such services outside the firm, a lawyer must make reasonable efforts to ensure that the services are provided in a manner that is compatible with the lawyer’s professional obligations.”

    SCR 20:5.3 requires the law firm to have in effect measures that give reasonable assurance that the assistance by a nonlawyer is compatible with the professional obligations of the lawyer and that a lawyer having direct supervisory authority over the nonlawyer assistance make reasonable efforts to ensure that the assistance is compatible with the professional obligations of the lawyer. In addition, the firm’s lawyers may be vicariously responsible for violation of the rules caused by the assistance.

    Ethical Issues Raised by Using ChatGPT

    To understand the ethical issues associated with using ChatGPT, I went directly to the source. Yes, I asked ChatGPT what legal ethical issues are raised by the use of ChatGPT. The following is its response.

    Response: There are a number of legal and ethical issues that are raised by the use of ChatGPT and other large language models. Some of these include:

    • Privacy concerns. ChatGPT and other language models are trained on large amounts of data, which can include sensitive information such as personal details and private conversations. This raises questions about how the data is collected, stored, and used, and who has access to it.

    • Bias and discrimination. Language models can perpetuate and amplify existing biases in the data they are trained on. This can lead to unfair and discriminatory outcomes, particularly for marginalized groups.

    • Misinformation and manipulation. Language models can generate highly convincing text, which can be used to spread misinformation or influence public opinion. There is also the potential for malicious actors to use these models to impersonate others or create fake content.

    • Autonomous decision-making. As language models become more advanced, there is the potential for them to be used in decision-making systems that have a significant impact on people’s lives. This raises ethical questions about accountability and transparency in these systems.

    • Intellectual property. As language models generate original text, there may be questions over who owns the rights to the generated text and whether it can be used for commercial purposes.

    It is important to keep these issues in mind when developing and deploying large language models and to have appropriate measures in place to mitigate any potential harm.

    Cite Use to Avoid Plagiarism Claims

    I followed up by asking if ChatGPT should be cited to avoid plagiarism claims. ChatGPT replied:

    Response: As a language model, ChatGPT generates text based on patterns it has learned from the data it was trained on. If text generated by ChatGPT is used in any form of publication, it should be cited to give credit to the model and the data it was trained on. Citing ChatGPT would also help to indicate that the text was generated by a machine and not written by a human, which may be important in certain contexts.

    I think it is also important to note that the generated text may include information from copyrighted sources, therefore, it is important to check the generated text for plagiarism and ensure that you have the right to use any copyrighted material.

    Citing to Authority

    The reply above prompted me to ask ChatGPT if it made up citations to authority. It responded in a lawyer-like fashion:

    Response: ChatGPT is a language model that generates text based on patterns it has learned from the data it has been trained on. It does not have the ability to create citations or reference sources on its own, but it can generate text that may contain citations or references if it has seen such examples during its training.

    However, I think it’s worth noting that the model is not able to verify the authenticity of the citation or the source it’s referencing, and the generated text might contain false or inaccurate citations that were present in the training data. It’s also possible that the citation may not be formatted correctly, or it might not match the citation style required for the context.

    It is important to be aware of these limitations and to verify any citations or references that are generated by ChatGPT, especially if they are being used in an academic, scientific, or professional context. Additionally, it’s important to note that the model can be fine-tuned on specific domain and it can be used to generate text with citation, but again, the veracity of the citation should be checked by the user.

    In summary, ChatGPT can generate text that contains citations, but it is not able to create or verify their authenticity. It is the responsibility of the user to verify any citations generated by ChatGPT.

    Implicated Rules of Professional Conduct

    As ChatGPT itself admits, its use raises several ethical concerns. I discuss a few of the Rules of Professional Conduct that are implicated below.

    • Responsibilities Regarding Nonlawyer Assistance. SCR 20:5.3 is an umbrella rule. It requires the law firm to have in effect policies and procedures that give reasonable assurance that the nonlawyer assistance is compatible with the professional obligations of the lawyer. The rule also requires the lawyer having direct supervisory authority over the nonlawyer assistance to make reasonable efforts to ensure that the assistance is compatible with the professional obligations of the lawyer. Failure to comply with this rule’s obligations is itself professional misconduct. ChatGPT acknowledges that it is important to have appropriate measures in place to mitigate any potential harm.

    • Confidentiality. SCR 20:1.6 prohibits the disclosure of all information relating to the representation of the client, whatever its source, unless the client gives informed consent to the disclosure or the disclosure is authorized by one of the specific exceptions. The prohibition also applies to “disclosures that do not in themselves reveal protected information but could reasonably lead to the discovery of such information by a third person.” ABA cmt. [4]. Whether a lawyer believes that a disclosure would be “harmless” is not relevant to the analysis of whether such a disclosure would be permissible.

      SCR 20:1.6 also requires a lawyer to act competently to safeguard information relating to the representation of a client against unauthorized access by third parties and against inadvertent or unauthorized disclosure by the lawyer or nonlawyer assistance participating in the representation of the client. ChatGPT admits that it is “trained on large amounts of data, which can include sensitive information such as personal details and private conversations.” It also acknowledges that “how the data is collected, stored, and used, and who has access to it raises confidentiality questions.”

    • Competence. SCR 20:1.1 requires the lawyer to provide competent representation to a client. “Competent representation requires the legal knowledge, skill, thoroughness and preparation reasonably necessary for the representation.” ABA comment [8] cautions that to “maintain the requisite knowledge and skill, a lawyer should keep abreast of changes in the law and its practice, including the benefits and risks associated with relevant technology.” ChatGPT reminds users that “it is the responsibility of the user to verify” any information generated by it.

    • Honesty. SCR 20:4.1, SCR 20:3.3, and SCR 20:8.4(c) prohibit a lawyer from making false statements. ChatGPT cautions “that the model is not able to verify the authenticity of the citation or the source it’s referencing, and the generated text might contain false or inaccurate citations that were present in the training data.” In addition, ChatGPT advises that “if text generated by ChatGPT is used in any form of publication, it should be cited to give credit to the model to indicate that the text was generated by a machine and not written by a human which may be important in certain contexts.” ChatGPT acknowledges that it “generates highly convincing text, which can be used to spread misinformation” and that there is “the potential for malicious actors to use these models to impersonate others or create fake content.”

    • Discrimination. Under SCR 20:8.4(i), it is professional misconduct for a lawyer to “harass a person on the basis of sex, race, age, creed, religion, color, national origin, disability, sexual preference or marital status in connection with the lawyer’s professional activities.” The Wisconsin Committee Comment following the rule provides guidance: “What constitutes harassment under paragraph (i) may be determined with reference to anti-discrimination legislation and interpretive case law.” ChatGPT acknowledges that it can perpetuate and amplify existing biases in the data on which it is trained, which “can lead to unfair and discriminatory outcomes.” In addition, SCR 20:4.4(a) prohibits a lawyer from using “means that have no substantial purpose other than to embarrass, delay, or burden a 3rd person.”

    • Independent Professional Judgment. SCR 20:2.1 requires a lawyer to exercise independent professional judgment in representing a client. ChatGPT acknowledges that there is the potential for it to be used in decision-making systems.

    Aviva Meridian Kaiser, Univ. of Buffalo 1979, is ethics counsel with the State Bar of Wisconsin. Ethics question? Call the Ethics Hotline at (608) 229-2017 or (800) 254-9154.

    Also of Interest

    Want to Learn More About ChatGPT?

    State Bar of Wisconsin PINNACLE® is offering “ChatGPT Artificial Intelligence – Overview and Implications” as a 1.0 CLE, live webcast seminar on Wednesday, March 15. Replay webcasts are scheduled for: March 28 and 31; April 6, 12, 18, and 24; May 30; and June 15.

    Visit Marketplace.wisbar.org for program and registration information.

    Endnotes

    1 See https://etedward-gptzero-main-zqgfwb.streamlit.app/;https://openai-openai-detector.hf.space/.

    2 Michael James Bommarito & Daniel Martin Katz, GPT Takes the Bar Exam (Dec. 29, 2022), https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4314839.

    3 Jenna Greene, Will ChatGPT Make Lawyers Obsolete? (Hint: Be Afraid) (Dec. 9, 2022), www.reuters.com/legal/transactional/will-chatgpt-make-lawyers-obsolete-hint-be-afraid-2022-12-09/.

    4 Bonnie Shucha, ChatGPT Chatbot Can Write Anything from Student Essays to Legal Briefs (Dec. 28, 2022), www.wislawnow.com/2022/12/chatgpt-chatbot-can-write-anything-from-student-essays-to-legal-briefs-but-author-and-reader-beware/.

    5 Id.

    6 Seesupra note 3.

    » Cite this article: 96 Wis. Law. 41-47 (February 2023).


Join the conversation! Log in to comment.

News & Pubs Search

-
Format: MM/DD/YYYY