chevron-down Created with Sketch Beta.

GPSolo Magazine

GPSolo September/October 2023: Protest or Riot: An Overview of Accountability

Use Artificial Intelligence Intelligently: Avoid Sanctions, Ditch the Billable Hour, and Become the Lawyer of the Future

Mathew Kerbis


  • The ability of generative artificial intelligence (GenAI) tools such as ChatGPT to accomplish legal work creates incentives for law firms to stop billing time and could completely disrupt the law firm model.
  • Solos and small firms, especially those using subscriptions and value-based billing rather than billing time, can be more nimble than larger firms and can sooner adopt GenAI legaltech tools.
  • The privacy and ethics concerns regarding AI are real, but attorneys can address these issues by carefully integrating the technology into their law practices.
Use Artificial Intelligence Intelligently: Avoid Sanctions, Ditch the Billable Hour, and Become the Lawyer of the Future
Westend61 via Getty Images

Jump to:

Technology specifically designed for law firm users (commonly known as legaltech) along with technology adjacent and adaptable to law practice has evolved rapidly since the pandemic, and even more so since the advent of ChatGPT in November 2022. Private practice firms are still not quick to adopt technology that will make them more efficient. Exercising some caution is a good thing because jurisdictions require technology competency in their codes of professional conduct. However, the hesitancy of law firms to adopt technology is not due to a lack of competency alone but also to a fear of a reduction in billable hours.

Builders of legaltech tools create products catered to the billable hour and contort their marketing to law firms on how their technology will not impact their bottom line. “Be more efficient so you can take on more clients” is a sales pitch often heard by firm decision-makers who begrudgingly meet with a legaltech vendor at the urging of a newer associate. There’s no question that there are ample opportunities to represent more clients if you complete client work faster, but the latent legal market, potentially valued in the billions if not trillions of dollars, doesn’t want to pay for billable hours. Those clients want pricing certainty and will expect their lawyers to use generative artificial intelligence (GenAI for short).

GenAI’s arrival, and the arrival of the pandemic before that, stimulated some attorney technology adoption out of necessity. Wide adoption of the cloud and remote court appearances only happened because the profession had no other choice than to adopt these modern technologies during the pandemic. Law firms did not adopt legaltech such as e-discovery until clients demanded it and because it became impractical for document review lawyers to analyze thousands, if not millions, of records. GenAI is a little different because there is widespread acknowledgment of the potential benefits to society and all industries, if not a widespread understanding of the right way to use it.

The infamous New York attorneys who unintelligently used ChatGPT, a free GenAI tool created by Open AI (a misnomer because its models are no longer open), to cite to cases that didn’t exist is a lesson in accountability for law practice management. GenAI can revolutionize the practice of law when used correctly. GenAI is like a toddler who has earned a master’s degree in English, has read everything on the Internet, is a Method actor, and has a photographic memory—but who can only access this perfect memory when prompted with questions (and who, after too much prompting, will even forget the previous questions asked earlier in the same conversation). Would you trust this toddler as a source of truth? Of course not! A toddler’s brain is not fully developed even if this toddler has perfect recall and can write in the style of any well-known author.

Like a toddler who can act, GenAI is surprisingly creative and has a great imagination—or at least something that mimics imagination. This GenAI toddler also wants to do anything to make its parents, or prompt engineers, happy, and it will make up facts using its clever imagination-like algorithm to seem really convincing. However, it’s going to be much better at helping you in other aspects of your law practice than providing you with facts or data. Instead, you must be the source of facts and data that you give to the GenAI and restrict its answers to the data you provide. Is it perfect? No. Is any piece of technology or any employee perfect? Of course not.

GenAI and the End of Billable Time

Those of us not billing our time found out quickly that GenAI has the potential to help attorneys scale their practice and serve more clients more effectively and efficiently without the need to hire more lawyers or staff. When understood and used appropriately, whether for solos or associates and staff, leveraging GenAI is like giving everyone a superpowered personal assistant. Everybody at a law firm, from the firm owner to the lowest paid staff person, should be using GenAI every day to help them work better and more efficiently at all times. Everyone gets an AI assistant now, even the human assistants. This technology will not replace anyone’s job as long as the firm uses GenAI to accomplish jobs that used to take several people without AI assistants constantly available. This is not unique to the legal profession but is true for all businesses.

Because they are not billing time, in-house teams and general counsel will likely be early adopters of legaltech tools using GenAI to help them reduce their demand for outside counsel. For this reason, many legaltech companies target in-house legal teams as their initial market. Similarly, solos and small firms, especially those using subscriptions and value-based billing rather than billing time, can be more nimble than larger firms and can sooner adopt GenAI legaltech tools appropriately priced for solo and small firms, allowing them to better serve the latent legal market.

The ability of these powerful tools to accomplish legal work not only creates incentives for law firms to stop billing time, but it also has the potential to completely disrupt the law firm model as we know it, which is largely built on billable hours. In the long term, we may even see the elimination of the BigLaw model for law firms. Associates at these firms are highly compensated because the work is terrible and unbearable—there are high billable hour requirements, horrible office politics favoring white men, no work-life balance, high stress, and a bevy of other things I could spend the rest of this article listing. These BigLaw associates will see that they can use modern and near-future legaltech and non-legaltech to automate their own practice to serve more clients using subscriptions and/or flat fees, earn a very good living, and all the while help more people and businesses than they ever could have done at a large firm.

GenAI’s tendency to incentivize more profitable billing practices will create transparency and fairness in billing arrangements with clients. As GenAI is well known by the masses and even used internally by traditional corporate clients of BigLaw, clients will demand that their lawyers use this ground-breaking technology, and so lawyers must use it to keep those clients. Some lawyers have been writing articles and posting on social media about how billable tasks that used to take them ten hours now take them ten minutes. Sure, lawyers can increase their billable rate by ten times or more to make up the difference, but clients are not going to want to spend $10,000 an hour for legal services, even if they actually end up spending only $1,000 and get the result much sooner. Lawyers must adopt subscription and value-based flat-fee pricing to survive in the world of GenAI. One of the inherent benefits of subscriptions and value-based billing to clients is that the price of legal services is known at the start of the engagement with the attorney, and it is predictable and transparent by design.

Part of the aversion by law firms to any new technology is not only that it will reduce billable hours but also that lawyers generally have a hard time learning the new skills required to use new technology. For law firms that have staff, this generally means the staff will use the new technology, and the attorneys will communicate with the staff. Because GenAI uses natural language, incorporating it into legaltech can render this extra step unnecessary. Lawyers can now “speak” directly to the technology. A non-legaltech example can be seen in how the automated workflow services Zapier and Make have incorporated GenAI into their platforms. Instead of forcing customers to spend hours learning how to use no-code tools and the jargon around them, these platforms have integrated GenAI to allow users to get the automations they want in minutes using natural language.

Ethics, Privacy, and GenAI

Based on what we now know about how large language models (LLMs) are trained, law firms can handle in fundamentally the same ways both the privacy and the ethics concerns inherent in how GenAI treats client data. One way is for law firms to use an LLM-in-a-box that is pre-trained generally (a foundational model) and then additionally trained on firm data, and that is usable only by employees at that law firm. Law firms that adopt a subscription model can have subscribers to their legal services get access to a GenAI chatbot trained on public or non-confidential legal data so that clients can ask the chatbot questions via prompts before communicating with a lawyer to go over a contract or legal situation; the lawyer then can use the results of that initial client-to-GenAI chatbot conversation as a jumping-off point in providing clients legal advice and/or services. However, this chatbot cannot be trained on or have access to confidential client information and contracts.

When documents do not contain confidential client information or will be filed with a court or otherwise become public record, the privacy concerns of using GenAI do not exist. Imagine having GenAI take on the persona of a judge who went to the same school as your judge, who was in private practice for the same number years in the same practice area before being appointed to the bench, who started in, say, traffic court before moving to chancery matters, and so on. You could query this GenAI judge persona on which arguments in your brief it finds more persuasive. You can prompt the GenAI to change all passive voice to active voice and otherwise summarize your lengthy argument into a single paragraph. Do you have a motion to dismiss that you are particularly proud of? Or an asset purchase agreement? Give it to the GenAI as an example and then prompt GenAI to write a version of that document in the same style but with new facts, cases, and data that you provide it. All this can be accomplished with tools currently available and does not even require that the GenAI tool be trained specifically for legal services.

There are some new legaltech GenAI tools that are trained on legal data and promise that they do not train AI models on your client or firm data. These tools are excellent assistants in summarizing large documents, extracting specific information, or comparing several documents. Could you throw several lawyers or even some existing technology at these situations to get the same results? Of course. But it will be expensive and time-consuming. While GenAI is prone to making up facts, it is less prone, in my experience, to making mistakes when performing these kinds of tasks.

Certainly, GenAI is less prone to making mistakes as compared to a new associate or staff person without a law degree. There is even a GenAI tool already trained on all U.S. federal and state statutes. You can ask it about any of those regulations or whether a document is in compliance with those regulations. You should still not exclusively rely on the answer, but you can rely on it more than you can rely on a less experienced legal professional who hasn’t memorized with perfect recall all federal and state laws. The natural result is that this kind of superpowered assistant will lead to the provision of legal services much more quickly, which means fewer billable hours.

The Future of GenAI

It’s difficult to predict how long it will take, but AI will be a non-negotiable tool used by all practicing attorneys in the near future. Could you imagine a lawyer not using a computer or smartphone today? Such lawyers are out there, but the quantity is negligible. The same will be true for attorneys not using AI. If attorneys want to stay in this business, they will be using AI and abandoning the billable hour in favor of subscriptions and value-based billing. For law firms, I think well-executed and transparent use of GenAI client-facing interfaces is a must. Execution is really important here, and existing chatbots are really an example of what not to do with this kind of technology.

GenAI has the potential to enhance the interaction between clients and law firms, and the intelligent implementation of such “attorney-adjacent” self-help GenAI chatbots can help provide immediate information (but not legal advice) to clients. There can be challenges in maintaining a personalized approach with clients depending on how a law firm implements GenAI. A personalized approach may not be necessary, given the price point of a particular legal service. Think about LegalZoom. That is not personalized at all, but it is a multi-billion-dollar publicly traded company. Being transparent that an interface is a GenAI chatbot trained by the law firm and on legal data specifically will go a long way to building trust with clients using that technology.

Subject matter expertise is still really important in this phase of how we interact with GenAI, and by that, I mean through prompting. As an attorney, I’m able to generate fantastic results from GenAI products via my prompts because I know how to give the right context to GenAI before giving it the right words to use in the right order, and when available, how to do so in a way that benefits from sequential prompting based on prior output. While I think I am creative, I am neither artistically creative nor knowledgeable, so in my attempts to use image-based GenAI, I get terrible results. For this reason, GenAI chatbots for legal clients will not replace lawyers anytime soon, and they should be trained not to give legal advice to clients who prompt them. It would be prudent for lawyers to give suggested prompts as part of the chatbot interface and to remind clients who use GenAI chatbots that this is not a substitute for legal advice, although it may be the next best thing in the meantime until legal advice can be provided to their situation.

It’s possible that the future of what a lawyer does will look more like either legal prompt engineering or using software with pre-built prompts that can be modulated by legal subject matter experts. In this future, not only will the business model of law be disrupted by the wide and necessary adoption of GenAI, but so, too, will the legal education system and attorney licensure. When GPT-4, a more advanced version of ChatGPT, passed the bar exam, it pulled back the curtain on the fiction that the bar exam is a reasonable barrier to entry for licensed attorneys. Instead, an undergraduate degree in law makes a lot of sense with where GenAI is taking us as a society.

I hope that this primer on how to think about using GenAI will help you not only avoid sanctions but also develop a practice prepared for the future.

This article is 100 percent human-generated.