As its widely known that Litigation Document Review takes up to around 60-75% of the cost of modern litigation. That makes us understand how important a precise DR is. But one thing that also comes with all the expenses is that DR is a time-consuming and a pain-staking process. To review a chunk of, let's say, 10,000 documents, it could easily take a team of 10 reviewers 4 to 6 weeks, taking into account the complexity of the documents, of course. This is where AI comes in.
Artificial Intelligence or AI, a product of science and logic, is an up and coming tool in the market, which helps people in doing things faster in many spheres of life. AI is helping Cyber Security, Medical / Life Sciences, Market Research, Accountancy, and of course Legal Services. There have been many tools in the market, which already had a big name in the market owing to their functionality and diverse features for DR are now trying to integrate AI in their workflow, with only one hope of cutting short the time and cost which a typical document review assignment yields.
Many people are still new to the concept of AI and are still under the impression that the integration of AI in DR is a concept in its nascent stage. Many organizations are working day and night to make this happen, such as Microsoft, Relativity, NexLP, Brainspace, etc. Artificial Intelligence combined with Machine Learning gives such results to the teams which help in reducing the time taken in reviewing all the documents.
AI integrated with DR works by making the machine learn on how the documents are to be coded as per the project protocol. There has to be a small number of documents coded in positive or negative binary values. For eg., a document could be either Responsive or Non-Responsive based on the protocol provided by the client. A set of such documents is given to the machine and the machine then looks for the subject-matter values in the documents which compelled the reviewers to make a particular choice for a particular document. The machine then returns the remaining documents in the set of Responsive to Non-Responsive arranged in the order of a Predictive Rank. The higher the predictive rank, the more chances a document has, to be coded as Responsive.
This helps the review teams in culling down the time taken in going through all the documents. With the documents now arranged in the order of their predictive responsiveness, they may now directly QC the documents and code them accordingly.
This method of indexing works on various types of documents such as invoices, email, purchase orders, policies, etc. Since processing and reviewing such documents currently is very time-consuming and prone to errors, AI is paving the way for a faster and more efficient Document Review. A study conducted by EY with Microsoft with such machine learning models has yielded surprising results by concluding the review 200 times faster than the current rate of review and the assessments made by the reviewers 10% more accurate than when humans were involved.
Organizations such as Microsoft with their Azure, Relativity with their Active Learning, and Brainspace with their CMML tools are trying to pave the way for human beings to conduct review faster than ever. Its time the world should shift towards Document Intelligence from Document Review.
It's safe to say that AI is paving the way for a better DR while the future is bright with all the document intelligence and AI tools, getting economical to use by the DR community. The legal fraternity would also be better off by saving some of their costs incurred in processing and DR.