---

Abhishek Mittal is the Vice President of Data Analytics and Operational Excellence for Wolters Kluwer’s Governance, Risk & Compliance or GRC Division

He directs the company’s advanced analytics efforts, including the AI at the heart of its products and solutions. In the Q&A below, he answers five pertinent questions at the very heart of the future of artificial intelligence and its role in legal technology. 

1. With respect to legal technology, what are the top three AI applications everyone should be familiar with right now? 

AI has been used across multiple areas in the legal industry. However, in my experience, three are being influenced by AI more than the others: Fact extraction, Pattern recognition and Predicting future performance. Concerning contracts, AI models are being used to extract critical facts for up to 70 data fields in a single contract, including party name, address, dates, amounts and many more. Also, the latest supervised machine learning models are quite good at using historical data to identify patterns across a high number of transactions and document types. An excellent example of this is from our AI-assisted bill review product.

In this solution, AI models can look at patterns at the timekeeper level, at the law firm level, and those patterns are used to come up with an overall compliance score off the invoice. Finally, AI technology is also being used to predict future performance correctly. Solutions are leveraging historical matter data to help a client budget for a new matter like those in the past. While AI is undoubtedly being used for other applications all the time, these are the three I’d say are most widely used now. 

2. Right now, what are the biggest concerns with using legal technology AI, and how are those addressed? 

While there can be a lot of worry about accuracy, machines do learn over time, and they do improve on their already high reliability. In my view, the more significant risk is not the AI or the machines getting it wrong – it is about what exactly the model is based upon and how that is being used. With respect to AI and its data, you must evaluate the overall quality and depth behind the solution. It is critical to determine the data’s integrity and how it has been labeled throughout the process. It is equally imperative to make sure there is a reliable and experienced team of experts behind the data.

Its depth is essential in terms of the number of records running through the model and the number of overall scenarios. If varied examples are not given to the model while training, it will not work in those scenarios that happen in production. Suppose you are using a model which does not have these kinds of checks done. In that case, there is a high likelihood that the model will not perform in real life, or worse, it might perform initially, but it loses its power over time. The most considerable risk of using AI is buying it without knowing what data it has been trained on and how much experience it has in real production scenarios.

Source: Freepik

3. How is AI-assisted legal invoice review better than non-AI-assisted? 

Machines are great at certain things, and humans, of course, have their expertise. When these two are combined, it will drive the most successful outcomes overall. There are three areas in which AI-assisted legal invoice review is better than a humans-only approach. First, the machines are better at identifying patterns across invoices and similar matters, but this process is much better when adding the human experience. Individually, both have their strengths, but having AI can prioritize and allocate the work according to expertise when you include both in a solution.

Also, machines can quickly identify a very high number of easy patterns, which leads to improved accuracy and improvements to current and future versions of AI-powered solutions. But for those areas based on a large amount of subjectivity, humans are better suited because they can leverage their institutional knowledge to get the best results. Another tremendous benefit of the AI-assisted solution – especially if your company has a large operations volume – is that you can drive much higher legal bill and invoice review consistency. AI creates that reliable baseline level of quality across a team based on varied experiences and skill levels. With AI-assisted solutions, you can provide consistent and high-quality outcomes.

4. How will AI-assisted legal invoice review become even more powerful in the future? 

We will continue to see improvements in the models and their accuracy overall. As the option of AI-assisted legal bill review and other services increase across industries, the models will just become more and more precise. I predict the bulk of invoice reviews will be done by machines as time progresses. But another development I feel is ahead of the game is that machines will become even more responsive in real-time. As the timekeepers make their entries, the devices will identify something immediately that is non-compliant get that corrected. 

Rather than being a back-end process, legal bill view - I feel with the advanced sophistication of algorithms concerning the data and the technology - will become a much more near real-time operation. As AI-assisted invoice review adoption grows across industries, AI models will become more accurate and conduct the bulk of the invoice review with near real-time responses and help timekeepers be compliant at the time of billing.

5. As AI technology improves and become more prominent over time, will AI-assisted bill review completely replace humans? 

Right now, AI technology is not close to replacing humans. Technology is just not advanced enough to take over all human job functions – especially concerning legal bill review. Even with as good as it is right now, the AI within all solutions still must be trained and fed information to work on very specific functions and parts of jobs.

Right now, AI works in certain critical aspects of these solutions because we can train them to look for certain patterns and the like. But because of the high level of subjectivity across all these solutions, humans will still very much have a role in working together with AI to get the best results for businesses.

At Wolters Kluwer, our solutions will continue to enable the experts to really showcase their knowledge and expertise. Right now, there are many experts across a wide range of industries and businesses that are still spending too much time doing the grunt work. When looking for the right company to provide your solutions, make sure you find one that offers a team of experts behind the technology to think of and solve the nuances and details that otherwise might get overlooked.