According to the Gartner Hype Cycles of the past years, artificial intelligence (AI) is one of the hottest emerging technologies of the moment. All over the world, governments are making multi-billion dollar investments to stimulate utilization, development and education in the field. However, within the legal profession, AI developments do not really seem to get through at the same pace. To fully understand why this is the case, it is important to understand the history and the basic concepts of this technology.
AI is about the development of computer systems mimicking skills normally requiring human intelligence. The beauty lies in the fact programmers don’t have to explicitly program rules in order for the system to have these skills; these are automatically extracted from historic data. The complexity and the volume of the data greatly influences how well these rules can be picked up from past events. More academic attention, computational power and an hugely increased data volume all contributed to the current glory days of AI.
Under the hood of every AI model, lies an algorithm which attempts to optimize a task. For example, the search engine of Google attempts to optimize the order of its search results, so that you find what you are looking for as fast as possible. Generally speaking, it does so very accurately. However, sometimes you need to alter your search terms in order to find what you are looking for. Technically, the model predicts what search results will be clicked on, based on the search terms entered. This implies by definition a probability that the prediction is not correct. In the setting of search results, this is rather harmless. Yet, if you consider that autonomous cars are actually continuously ‘predicting’ whether the wheel should be turned, or the brake should be pressed, are you then willing to get into that car?
A comparable discomfort is felt when discussing artificial intelligence in the legal profession; how can you trust such systems if the system by nature can almost never be 100% sure about its predictions? This brings us to another weak point of AI: algorithm transparency. Especially with advanced models, it is impossible to trace back how a prediction came to life as the logic is based on big matrices of decimal numbers being multiplied and mixed with each other to eventually come to a statistically optimal generalized result. However, in the legal world, where having a proper audit trail is of major importance, approaching problems probabilistically feels very unnatural.
In the legal world, where having a proper audit trail is of major importance, approaching problems probabilistically feels very unnatural.
Another problem why AI is only accepted gradually in the legal profession is the nature of the job; the legal profession is a rather textual world. Regardless of what fancy commercials like to propagate about computers being able to comprehend text, is simply not true yet. Probably the most active field in AI research is making computers read and understand text. The results are promising, but often not good enough for professional usage.
Finally, within organizations, the IT and the legal departments are often very separated, sharing little knowledge. As a lot of business understanding is required for a successful technology project, the chances of finding those in the legal department are relatively low. By teaching legal professionals more IT skills, they are better capable of spotting opportunities where the short comings of AI are less important towards the business.
By teaching legal professionals more IT skills, they are better capable of spotting opportunities where the short comings of AI are less important towards the business.
Fortunately, initiatives focused on sharing knowledge and skills between the legal profession and IT are popping up everywhere, crossing borders of traditional published media to get the message across. By experimentation in the form of meetups, blogs, podcasts, online courses, conferences and network groups, we see the two industries gradually merging together. The coming years we will notice many more advancements in both the knowledge sharing and the actual developments of applications within the LegalTech field. In my next blog I’ll share some specific areas of the legal profession in which I see short and long term opportunities.