# Major Law Firm Partner has concerns about AI use at their firm! A Major Law Firm implemented AI across all departments and discovered that absent specific utilization intentions, trust in the firms work product fell. A major law firm last year implemented at their firm, one of the most powerful AI solutions in the legal field. This implementation was done last year. According to Brendan Dell, a partner at the firm intimated to Brendan that the firms employees do not trust each others work any longer. As he explains in his video (link below), the Associates doing the work don't trust the AI output, so they triple check every thing. The Managers, knowing that the Associates are using "the AI", don't trust the work produced by the Associates, so they triple check the work provided by the Associates. Of course, the Partners, who review the Managers work before it goes out to the clients, are concerned about the AI input that the Managers may have utilized. ![Legal Team Members dont trust the work produced by colleagues when AI is used.](https://cdn.sanity.io/images/hxpc8b0a/agents/5758a812b931fe379eb7932e98af9e54fbbaa449-1897x853.png?w=1600&fm=webp&q=80&auto=format) _Partners triple check the work done by Managers. Managers triple check Associates work._ This leads to the cascading effect that Partners are now triple checking the work produced by the Managers. Exactly where is the productivity gain here when the distrust of the AI Output is so high? In another example provided by Mr. Dell, he cites that Insurance Industry professionals, in their haste to get AI into the production work of Claims Processing, are finding enormous errors. To better illustrate the type of errors they are getting, here is a post that Mr. Dell shared in his video (link below). ![Insurance Carriers using AI for claims are seeing major productivity losses](https://cdn.sanity.io/images/hxpc8b0a/agents/ccbc3831dd733fc9944709f8a3a3bbe3f40869e6-1000x267.png?w=1600&fm=webp&q=80&auto=format) _The LLM keeps failing to differentiate between zero and "o"._ What Firm and Corporate Leaders should be looking at before they implement an AI solution enterprise wide is: - What specific tasks do we want less friction where an AI could help with? - What are the limits of AI tools in solving for less friction with that task? - What specific, measurable benefits will AI bring to the task? - Lastly, what will the "tokens" cost be for having AI do this task? These major questions all stem from a quiet but enormous problem that most firms and businesses don't look at directly. That question is, "what does each person do" with respect to a task that we are looking to solve with AI Automation. To provide an example of our meaning, it is very common when a human trains another human to do a job, that many steps are left out of the training process. These steps are often 'interpretive' steps where the experienced employee knows what to do if "this specific thing" happens. Of course, there are many things that are known by management, when looking at a process that hasn't been documented extensively. Because they are not known by the business or firm, but only by the employee or the staff, they are rarely documented. Everyone on the team knows that Elaine doesn't arrive at the office until 9:45 AM on Thursday's. So, on Thursday's, all the initial scans that piled up during the night are given to Jim, who divides them up between the interns and the temps, depending on their workloads. If the AI is going to automate a process, and Elaine provides a role prior to AI's task list, or post Elaine's work completion, what happens on Thursdays if it isn't documented? The final element that the Firm or Business leadership must have a solid handle on, is what exact data is the AI Agent learning from? If it is learning from the internet, then there is a high likelihood that for specific business or firm processes, there will be errors. If the Agent is learning from well configured data, that is stored in a best practice architecture, and is focused on the internal data of the firm or business, that is specifically relevant for the task at hand, better results will follow. Sometimes being first to market is the single most important aspect of why a business or firm is successful. This includes the use of technological innovation. Heinz Corporation took full advantage of the Electricity Boom before any other major manufacturer, and as a result became the largest food producer in the world. But, in the case of Artificial Intelligence, the cost of a token (how you pay for AI use to the service providers) is not fixed or expected to go down in the future. And, in this case, we dont have the same kind of understanding of what the AI is replacing as we did when we switched from fire to electricity to light a room. What we mean is that employees are able to navigate a complex set of circumstances in seconds, to determine and decide, the best plan of action for a given task. An AI will need a map for nearly every possible pathway that a given workflow can take, before it will be able to statistically determine what may be the path to take. AI is here to stay, and firms and businesses will benefit from its use. For that future of benefit to come to pass, a little more measuring before any cutting may be what the prudent do. To watch the Youtube Video of Brendan Dell, from where he talks about Law Firms, [click this link](https://youtu.be/J6J8vhIzfLo?si=F918x-XJ0dS7zqBl). If you want to watch it from the beginning, [click this link](https://youtu.be/J6J8vhIzfLo?si=eqMRK-ya2Yj-ooLk).