Not long ago my 14-year-old son was complaining about the fact that his ten-year-old sister had just gotten an iPod touch. His objection? He hadn’t gotten his first iPod until he was 12. I responded, “When your six-year-old sister turns 12, she will have devices you haven’t even heard of yet.”
There is a Persian tale about the inventor of chess presenting his invention to the king, and the king was so pleased he allowed the inventor to name his award for the invention. The man asked the king that on the first square of the chessboard he would receive one grain of wheat, two on the second, four on the third, and so on, doubling the amount each time. The king quickly accepted the offer, thinking it was a low price to pay. However, when the amount was calculated (264 – 1 or almost 78 million tons of wheat), the king discovered it would take more than all his wealth to provide the award.
In 1965 Gordon E. Moore, one of the Intel founders, wrote a paper observing that the number of transistors on integrated circuits doubles approximately every two years. This is known in the computing world as Moore’s Law. As it turned out, Moore’s Law became a self-fulfilling prophecy—partly because the observation was based on empirical data at the time, but more importantly because the law is now used in the semiconductor industry to guide long-term planning and to set targets for research and development.
While the technology around the legal world advances at an exponential rate, the technology within the legal world, especially as it relates to lawyering (i.e., providing legal services as opposed to running a law business), is much slower. There are good reasons for this. Lawyers in general are risk averse. We need to be. One of the primary benefits of using a lawyer for legal services is to obtain some level of guarantee that the advice or outcome of our services will actually accomplish the purpose for which the services were provided. This places a heavy burden of responsibility on lawyers to ensure not only that the actual services they provide but also the manner in which they provide these services will not in some way increase the risk of breaching this important obligation.
For each new technological advance, a high level of analysis and review is needed before lawyers can implement it. Our professional obligations demand that we not take risks, so we tend to stick with what we know works. This creates tension between the risk aversion in the legal profession and the ever-changing expectations and demands of legal service consumers. This tension is measured as the difference in the rate of adoption of new technology by legal service consumers and the rate of adoption of new technology by legal service providers. Assuming Moore’s Law represents the rate of technological advance for legal services consumers (i.e., doubling every two years), and assuming a slower rate of growth for legal service providers (perhaps increasing by one-half every two years?), then every two years the tension grows exponentially at a rate of 1.5 times. As that tension continues to grow, the legal services industry will find itself competing with outside providers attempting to fill the gap. This phenomenon is already taking place and, as legal service providers, we should not ignore it.
Artificial Intelligence and Law
Merriam-Webster defines artificial intelligence as “an area of computer science that deals with giving machines the ability to seem like they have human intelligence” and “the power of a machine to copy intelligent human behavior.” As a technology enthusiast, I spent much of my time in law school and subsequently in my career exploring the application of artificial intelligence to the analytical skills needed for successful lawyering. I am convinced that although “machines” may assist in significant and meaningful ways, they are not yet (and I am doubtful if they will ever be) able to fully replace the value of skilled lawyer involvement in the provision of legal services. Having said that, I admit that there are many areas of legal services lawyers have traditionally provided that a machine can very easily and will eventually replace.
Electronic discovery (e-discovery) is an area where there are numerous examples and implementations of artificial intelligence both replacing work product from lawyers and at the same time improving the experience of law practice. Much has been written, discussed, and litigated around the technology and benefit of e-discovery. The focus of most providers is to provide intelligent algorithms to find information based on concepts and key words agreed upon by the parties to the litigation. The “hot new thing” in e-discovery is the use of a higher-level artificial intelligence concept called predictive coding.
Essentially, predictive coding is a process whereby a machine learns from watching human behavior and then applies what it learns. This is the technology behind how Amazon and Google seem to always know what you are looking for before you start looking. The machine’s learning algorithms are designed to gather data, analyze it, and then make decisions about what is relevant. And because of the increased computing power on these machines, this is done very quickly. A recent survey of Fortune 1000 Counsel by FTI Consulting indicated that 57 percent of their respondents believe that predictive coding will be a mainstream tactic by 2015. These respondents also indicated that in addition to technical aptitude, successful predictive coding e-discovery requires skills in statistics, accounting, project management, and linguistics.
If we apply Moore’s Law to e-discovery, however, predictive coding will be the old technology in 2020. So what comes after? From my research, I believe the most likely next step is the use of inductive logic programming. This is essentially where a neural network is created, and with minimal supervision the network can use speech and images to detect patterns or objects. In 2012 Google’s X laboratory created an incredibly large neural network of 16,000 processors and turned it loose by giving it access to 10 million YouTube videos. On its own, the network taught itself to look for cats. The concept of inductive logic programming is not new, but with the falling costs and increased power of computing capability, it is now something that computer scientists can start implementing. It’s just a matter of time before it makes its way into e-discovery tools.
Changing the Way We Practice
The key to our future success as legal service providers lies in our ability to identify the specific lawyering areas in which we can be replaced and those in which we cannot be replaced. The most prosperous law practices in 2020 will be those that are able to successfully adjust their business models to use artificial intelligence–type tools while at the same time promoting and delivering the part of the legal service value proposition that the machines are not able to provide.
Consider for a moment the success of non-lawyer legal service providers such as Rocket Lawyer and LegalZoom. Both of these online services provide the ability for legal service consumers to create their own legal documents and forms. Both services proclaim that they do not provide legal representation, are not law firms, and are not a substitute for an attorney or law firm. Yet they essentially provide the same deliverable that many of us do: contracts, wills, business formation documents, bankruptcy filings, and the list goes on. They provide consumers with the ability to create documents that are intended to accomplish a specific legal purpose.
These services have identified and are executing a business model in areas in which artificial intelligence can replace lawyers. They also recognize that there is a significant limitation to the value of these services. At the end of the day, none of these services can guarantee that the deliverables they are providing can and will accomplish the purpose for which the consumer obtained that deliverable. In fact, while their marketing efforts portray the results of their services in a positive light, they clearly disclaim any liability from attorney-client privilege and legal work product. Understanding this limitation in the product they provide, these services in almost all cases now also provide services to connect their consumers to lawyers.
When legal service consumers hire a lawyer to create a legal document for them, they actually receive two items of value: first, a document or other work product that meets the legal needs of the consumer (the more “replaceable” part), and second, some level of guarantee of the result by the attorney, which only a competent attorney can provide, with malpractice insurance and a structure of professional responsibility requirements as supporting elements of the guarantee if something goes wrong (the more “irreplaceable” part). The most successful law practices in 2020 will have found a better balance between minimizing the cost of the replaceable parts for legal service consumers while maximizing the cost of the irreplaceable part.
As we consider these concepts, the outstanding question is whether or not these non-lawyer legal services should be considered an enemy or an ally.
The Billable Hour
There is another significant factor that adds to the aversion of technology adoption for lawyers. While many firms are exploring fixed fee and unbundled legal service models, the billable hour continues to be the most prominent revenue model for lawyers. Under this model there is little motivation to spend “unpaid” time innovating, learning new technologies, or implementing tools that create lawyering efficiency.
Several years ago a partner in a firm I worked for asked me to “automate” a lease he used for a large client who owned high-rise office buildings. The lease template was about 65 pages. It took me only a few hours to create an automated template. Using the template, it would take him only about 15 minutes to generate a first draft of the lease for new prospective tenants, whereas it typically took him about three or four hours of billable time to generate the lease manually. As I presented this automated template to the partner, he was very pleased with the results at first. Several minutes into my presentation, however, he asked the question that doomed the project: “How will I make any money from this?”
The partner earned thousands of dollars in revenue based on the time it took to create the initial draft of the lease. He also earned revenue on the subsequent hours spent negotiating the final draft. With the automated solution, though, it would take only about 15 minutes or less to create the first draft. So based on a billable-hour model, his revenue expectations for this client would be cut nearly in half. I suggested that he consider changing his revenue model to charge a fixed fee for the creation of the lease and charge by the hour for any negotiation and modifications to that lease. He didn’t feel that the law firm would support this revenue model and decided not to implement the automated solution.
Lawyering Technologies in 2020
Over the next six years, the available computing power will likely double at least twice and maybe three times. This increase in computing power and new devices will likely drive current lawyering technologies such as document automation, decisions engines, e-discovery tools, communication and collaboration tools, legal research tools, and legal expert systems to continue to mature and progress in functionality and availability. Most of these tools started out in a desktop delivery environment, but there has been a sharp increase in the number of online tools available to attorneys. This trend will continue so that by 2020 most of the viable solutions will be available either exclusively over the Internet or with very limited desktop interfaces. Technology service providers are becoming device/operating system/browser agnostic. Trending industry concepts such as “big data” and “unstructured databases” will allow vendors to provide more robust, higher performance, and increasingly feature-rich applications.
Other technology trends will allow solo and small firm attorneys to access more information less expensively, to create and join communities of attorneys, and to collaborate within these communities in a way that is similar to having the networking and support that attorneys enjoy in large firms. The net effect of these trends will empower these attorneys to compete at a higher level with larger firms.
Perhaps most important to the future of the law practice, we will also see a continued influx of non-lawyer service providers enter the legal market, some of which will be exclusively consumer focused, some lawyer focused, and others will sell their wares to both consumers and lawyers.
By 2020 the delineation between legal service deliverables (i.e., work product and the value of the lawyer stamp of approval on that product) will be much clearer than it is now. The question remains, will the artificial intelligence and technology that drives the creation and delivery of the work product be coming from the law office or somewhere else? I believe the answer to that question depends on us, the lawyers.