Digital White Papers

I L T A S U R V E Y R E S U L T S | A R T I F I C I A L I N T E L L I G E N C E & M A C H I N E L E A R N I N G 4 C O M M E N T A R Y R E L A T E D T O T H E Q U E S T I O N S : Of the people developing or purchasing AI tools, most of them are purchasing out-of-the-box tools or customizing out-of-the-box tools. Only a few firms have the internal data science and/or development teams who can develop their own tools. There's an even split for where the data and legal tool resides or will reside - on prem or SaaS or cloud- hosted. That result mirrors the legal industry generally at this point, an increasing number of organizations moving to SaaS and cloud-hosted options for new technologies and applications. Of the organizations developing or purchasing a tool, there are some looking to use open source, Azure Cognitive Services and IBM Watson, while the majority are using whatever internal tool they already have in place or new and less well-known tools. How broadly legal AI-powered software is in use by organizations varies significantly, with one third saying they haven't figured that out yet to one third saying they plan to use AI tools across multiple departments, to under 10% saying they plan to roll something out globally. As the use of AI tools increases, these numbers will probably change substantially. Interesting to see who is or will be involved in any AI-powered software purchase, development, training and deployment – The CIO and IT is a part of at least 80% of projects, with attorneys next at 60%, innovation teams at 40%, then KM and legal operations teams and project managers. Not surprisingly, data science teams are involved in about 10% of projects, only those where an organization has data scientists. Likely others outsource or expect their IT to perform those duties. Data not cleansed or normalized is seen as the biggest challenge with the available current data set for the AI- powered software being developed or purchased. The other big challenge is insufficient data. This challenge is one we all need to focus us on to help mature the tools. The biggest challenges seen with training AI-powered software are analysis time for training and validation of results, unavailability of skilled personnel to do the training, inability to get good training data and data viability for usage. Good training is absolutely vital to successful AI projects, so working through these challenges will be the key. As with all IT Projects, good quality assurance can mean the success or failure. Requirements Analysis may be a bit harder to do for AI projects but it's listed in the top two QA elements incorporated in the reported AI projects, along with running a full Proof of Concept (POC). Validation documentation and documented success metrics are also seen as required to make these AI projects successful. A smaller but still significant number of organizations are running dual projects, with a full manual process running alongside an AI tool. That concept works for some efficiency tools but isn't as feasible for predictive tools. Mutual collaboration between law firms and law departments was seen as hugely important for success in AI projects.

• Cover