A common problem in contracting is that despite best intentions, contracts often don’t deliver their full intended benefits: according to the Harvard Business Review, enterprises typically lose anywhere from five to 40 percent of the value of any given deal over the course of its lifespan.
Such value erosion can have many culprits: suboptimal and/or inappropriate terms and conditions, failures by one or more parties to abide by their contracted obligations, and the absence of or poor adherence to best practices for contract management are among the common themes that prevent businesses from realizing the full value of their agreements. But technology is providing enterprises with newer and more sophisticated tools than ever before that can help reduce such problems – and eliminate arduous, manual aspects of managing contracts and conducting contract reviews at the same time. One such innovation, machine learning, offers tremendous promise in this regard.
What Is Machine Learning?
A branch of artificial intelligence, or AI, machine learning is when computers generate new algorithms and models for transforming data into business insights. Essentially, computers use the information they are given to teach themselves new tasks and applications, instead of being specifically coded to perform these actions.
Machine learning creates a feedback loop where as a computer is presented with greater amounts of data and types of tasks, it uses the results and learnings to refine its actions and to discover and perform new ones. There are several categories of machine learning based on the underlying data. These include:
Supervised learning, where computers are provided with data inputs that have been pre-mapped and labeled by humans;
Unsupervised learning, where the data being used is unlabeled and the task of discovering and mapping the information is automated (i.e. performed by a computer);
Semi-supervised learning is a combination of supervised and unsupervised learning, using data that contains some labels or specified fields and other blocks of information that must be crawled and classified by a machine.
How Does Machine Learning Differ from Artificial Intelligence?
You might have heard the terms machine learning and artificial intelligence used interchangeably, but they are not one and the same. Machine learning is just one type of artificial intelligence, but AI refers to the entire spectrum of intelligent computers and computer functions, including voice-controlled devices, robotics, process automation and natural language processing, which allows computers to understand instructions in human languages instead of computer code.
Many new and evolving technologies incorporate multiple types of AI. For example, a self-driving car leverages robotics, visual, natural language processing, machine learning, and other types of AI in order to recognize, process, and respond to real-time data inputs it receives on the road.
How Will Machine Learning Help Improve Contract Review?
If you’re reading this blog, you’re probably more interested in how machine learning can assist with the key challenges of contract management than in self-driving cars. Fortunately, machine learning offers some highly relevant and useful applications for contract review. Some of these already exist and some will allow for ongoing improvements to this function in the coming years.
There are several key points within a contract lifecycle where it is important to closely review agreements. Prior to a contract being signed, it is critical that legal, procurement, and all other key stakeholders review the terms of the deal to ensure they are feasible and that contractual risks are addressed and mitigated to the greatest possible extent. After a contract is signed, additional reviews are required to inform an implementation plan, and to ensure and measure contract performance.
Currently, the most useful machine learning-based applications for contract review assist with the tagging and classification of data. One of the most onerous manual tasks of contract management, this typically requires contract managers to pour over pages and pages of contract documents in order to extract pertinent information and contract milestones as part of the onboarding process for executed documents. Automation, another AI subset, allows computers to perform this work, while machine learning helps those computers to learn from every document they process, improving and refining future outputs.
Moving forward, machine learning will allow machines to better correlate contract terms with results and then suggest optimal contract inclusions or refine existing language. During contract reviews, this will help stakeholders to more easily identify and address contract-related risks.
What You Can Do Now to Prepare for Machine Learning (and Other Data-driven Innovations in Contract Management)
Machine learning requires volumes of digitized data from which computers can learn. As such, to benefit from both existing and future applications of machine learning, getting contract-related data into a ready state is key. For starters, digitizing all contracts and uploading them to an electronic repository will allow you to run machine learning applications. This includes ensuring that all contracts are classified and tagged appropriately – though as discussed, machine learning is already making this requirement easier to fulfill. Standardizing contracts – so that your agreements conform to similar formats, taxonomies, and use standard labels – can also help to prepare you for machine learning by making your data easier for computers to parse and understand.