Among the Knowledge Management (KM) community, one question we commonly hear is this: “How can I use artificial intelligence (AI) to populate KM repositories?” I have good news and less good news. There are pockets of success out there, but you need to be realistic about what AI excels at today.
KM can involve significant manual effort that entails subject matter expert involvement. Developing forms banks, practice-specific content, and matter and experience databases requires someone familiar with the area of law to parse through documents and matter information to identify standard practices and relevant data points. Because our subject matter experts are often timekeepers or otherwise have limited bandwidth, initiatives that require their input can often stall.
Not surprisingly, as technology advances, inquiring minds want to know if AI can replace some of this manual effort. The nirvana that many want to attain is for technology to successfully analyze piles of documents and data to build gold standard documents and provide insights into our matters and people. Complete nirvana is still a dream—or at least a lot more work than we imagine it to be. Nonetheless, creative KM professionals are coming up with ways to use today’s technology to augment the work of the KM team for the benefit of the practices. Note that this post is not discussing attorneys using AI to supplement their client work, for example, to aid in research and document review (technologies that KM may manage), but rather KM professionals or attorneys using AI to supplement their work in developing KM content.
At ILTACON 2019 in August, a fantastic panel presented on “AI-Powered Knowledge Management,” which directly addressed this question. Click here to listen to the recording. If you are interested in this topic, it is well worth an hour of your time. You can also read a recap of this session from the Legal Executive Institute here.
The four panelists presented on efforts at their firms to use AI to further KM projects. A common thread among their projects is that they are all using AI technologies to extract data and clauses from final documents with software often used for contract analytics, such as ContraxSuite, Kira or iManage RAVN Extract, which then populates a database for easy reference and/or further analysis. Stated differently, the panelists are applying the technology to unstructured documents and pulling structured data points, such as dates, amounts, and specific clauses, from them into a database. Use cases were largely around collecting data from transaction documents for two main purposes: (i) to get a big picture of trends and (ii) to be able to find a matter in which a specific set of circumstances occurred.
Interestingly, most of these projects are still in pilot or limited production, validating that these types of use cases are still relatively new. A KM peer I interviewed for this blog post noted that the relevant technologies have been used by law firms for only a few years, and the early days were likely spent on rolling them out for their core purpose, typically contract review for due diligence. Firms may only recently be exploring other applications for these technologies.
The panelists were united in their lessons learned. These projects require investment—from up-front project design and model training, to subject matter expert validation, augmentation, and analysis of results. Like with many other applications of AI, technology is streamlining the work, but not replacing it. Someone must dedicate time to and prioritize the effort. But once you get started and become more familiar with the capabilities of AI, you will think of more use cases that leverage the ability to extract data points. Extracting data alone is useful, but coupled with metadata tagging and data visualization tools, such as Power BI, it can drive powerful insights—and attorneys get very excited about these insights.
My nirvana state of using AI to build gold standard documents is not quite there yet—at least, not at a level of accessibility to apply broadly. If you have a very specific use case and the volume to justify the work, you can combine professional services, technology, and a lot of elbow grease to achieve this goal. But for most use cases, the level of investment required today may be too great for the payoff.
However, if we focus on using AI to supplement and streamline steps in the process, rather than rely on it to achieve the end goal, it is clear that AI can contribute toward document repositories as well as databases of information. One way to employ AI to create KM document collections is to use contract analytics software to isolate provisions in work product that require redacting, so you can add sanitized precedents to your forms bank. Another is to extract similar clauses across sets of documents to speed up the work in analyzing standard, alternate, and nonstandard language. The challenge here is crafting a human-technology partnership that leverages the abilities of each in an efficient manner.
Technology is changing rapidly and may lead to different possibilities within the next few years. In the meantime, though, AI can be used to further KM projects provided that you set realistic expectations and design your project to take advantage of its current capabilities.#KnowledgeManagementandSearch