We have now covered the basic precepts of Machine Learning. In order to really understand the process, however, we need to delve deeper. We’ve addressed how machine learning works, but not yet why it’s important to decision-making.
Human elements almost always impact decision-making. It is natural to allow our personal biases to influence and dictate decisions we make both personally and professionally. These biases, however, can have unintended effects if not accounted for. Mistakes and imprecision can occur despite the best intentions. A new wave of legal tech is changing the way firms conduct basic decision-making. Services like our own LexSemble seek to minimize risk by erasing these potential flaws and creating a system wherein data is collected, measured, and used in ways that reduce the power of an individual’s or group’s biases.
Cognitive bias has been exhaustively studied in the fields of psychology and economics. Some of the most powerful notions we live by can be explained by these biases, such as Confirmation Bias, the Dunning-Kruger Effect, the Framing Effect, the Status Quo Bias, and the legendary clash of correlation versus causation.
Let’s take one cognitive bias to use as an example. Law firms frequently rely on decisions made by people with specialized expertise. The so-called Cult of the Expert is a mainstay of decision-making throughout the economy, but it is intrinsically flawed. Let’s say that a legal team is made up of 5 people, but one member of the team is considered more expert on the specifics of a case. That team will likely defer to the perceived expert, rather than coming together and reaching a consensus based on multiple points of view. If one reads the definition of the Dunning-Kruger Effect mentioned above, the potential logical fallout is obvious. Things can go awry when one individual is given too much decision-making power, yet this is frequently how firms are run.
Rise of Machine Learning
Machine learning is at the forefront of a sea change currently underway in law. It cuts through the fog of human bias in favor of analyzing raw data. AI programs don’t have any intrinsic bias, and as long as they are trained and well-generalized, these programs can streamline decision-making in virtually any arena.
Machine learning does not have to replace human decision-making entirely. The robot apocalypse is (probably) many years from now. But machine learning is a powerful agent that can get in the middle of decision-making to bolster the checks and balances of a law firm and contribute the most thorough analysis possible. Next week, we will delve even deeper into crowd-sourcing, and provide some concrete conceptual details on the machine learning process.
This article is part of our 7-part Intro series. The others can be found here.