Switzerland: User Adoption as a Challenge in the Use of AI

We now have powerful language models and a wide range of AI tools. The legal requirements can also be mastered to a greater or lesser extent. But how do we ensure that users also use the possibilities of AI to their advantage? To deal with this challenge, we have published a “Use Case Guide” today, which we use internally in order to achieve more quality and efficiency in our work every day with smaller and larger steps. Part 29 of our blog series on the responsible use of AI in companies.

The hype around the use of generative AI has evaporated in many places. Some companies have improved customer service with chatbots on their website, others have implemented internal projects (and many have rehired) and most companies are desperately trying to provide their employees with some kind of AI tool. They gratefully fall back on offers such as Copilot for M365 from Microsoft, which are literally forced on them. So they can at least say that they also use AI. However, we constantly hear from disappointed users who are not really satisfied. The danger: Users lose interest in using AI or use private, uncontrolled tools – many, for example, find ChatGPT much better than Copilot.

Because, in our opinion, most of these tools are far too expensive if they are used across the board, and on top of that in the vast majority of cases are neither functional, integration nor legally satisfactory, we have dispensed with them. A few exceptions prove the rule. As is well known, we have developed our own solution for our needs – and this is also gaining more and more popularity outside the law firm (it can be used by anyone). We also have professional secrecy under control.

Challenge No. 1: We are creatures of habit

So we have solved the technical, legal and cost aspects. However, we are also struggling with two other challenges. The first challenge is that we are creatures of habit. We do our work in a certain way and are often either unable or unwilling to adapt it – even if it would make sense. This familiar “way” typically does not include the use of AI. The result is that while we find AI and its possibilities fascinating and don’t want to miss anything, we are very bad at incorporating it into our processes and rebuilding these processes if necessary. Very often, however, we simply lack the thought of transferring a certain step in our everyday office life to AI. So opportunities to improve the quality or efficiency (or productivity, to put it another way) of our work slip through our fingers.

Some examples:

• I need to fill out an Excel form with the details of a case. I could have the AI do this, but it didn’t even occur to me at first that it could do that with our tool – even though I programmed it myself. It does in two minutes what might take me an hour. I still have to check the result, but I’m definitely faster, and often better.
• I write a memorandum and know what I am writing. But with AI, I suddenly get a different perspective: an “advocatus diaboli” who can pick apart my arguments within seconds. Not that everything the AI writes has hand and foot. But it does make my job better. And in the end, she writes me the executive summary within a minute. However, I have to remember to ask them to do it, and I have to learn how best to do it. The good use of AI is also a skill, and even I, who know the technology inside out, am constantly learning something new here.
• I check a contract, jump back and forth, look for a job I’ve already seen somewhere up – and sometimes lose a lot of time with it. The AI can’t check the contract for me the way I can, but it is at my side in the form of a chatbot and does all the little things that help me. However, I first have to learn to deal with the fact that I always have a willing “AI buddy” at my side. I’ve never worked like that before. In the meantime, I always let the AI give me the “big picture” first, which makes my review even easier. Our solution delivers this in a few seconds, directly within Word – a decisive factor, as I have noticed.
• I was always used to looking for information on the Internet myself. I know very well how to do this, and Google knows a lot. But why don’t I use an AI research assistant that reads 50 websites for me in one minute and then delivers a report to me? So I have to learn to delegate (for example, how to formulate the order) and to control instead of doing everything myself.

So it’s a matter of every employee having to think about what processes their work consists of and where and how AI can help them. These thoughts do not arise by themselves, and often it will only be supposedly small things that the AI can do better or faster. But they add up. Why shouldn’t the AI also be allowed to check at the end of a contract negotiation whether all defined terms are actually defined in the contract? Or whether the redacted documents really no longer contain any personal data?

To provide an impetus here, we have started to compile our use cases – such as the examples described above – i.e. to show our employees where and how AI can help them in their everyday work. The result is a handbook with over 70 such use cases, and more are being added all the time. They are all implemented with our own tool and can therefore be implemented in your own company practically free of charge.