Wolters Kluwer approach to developing trusted AI

Verified content is key

First of all, we took a principles-first approach to AI. So not just technology led. We focus on our key strengths, which is our expert, curated verified content, and our network of subject matter experts and domain experts across a number of different fields. That is key to getting, AI right – more so than just the technology.

As an example; we have 7,500+ clinicians that work on our UpToDate product. We are really bringing together the best possible medical evidence to help doctors diagnose and treat patients.

Delivering AI you can trust!

Next up: Agentic AI in the workflow…

The next phase is Agentic AI, or using Gen AI in the workflow, focused on jobs to be done, reducing toil, and administrative burden. Here’s a good example in hospitals. Our UpToDate Expert AI solution is able to have a broader impact across the ecosystem in terms of workflow, reducing the administrative burden, and reducing burnout.

We're also very excited about that within Legal and Tax and Finance where we see other opportunities to automate workflows, always with the expert and human in the loop that is going to unlock more value from.

Not only the technology itself, but also the things that we've been doing for a very long time – which is creating expert, verified content, and then allowing the LLMs to bring that into the workflow, and getting in additional efficiency and impact for our customer. That's what we're going to be focused on next.

Trusted and verified content with UpToDate ExpertAI

Trusted AI is first and foremost. The key, the main focus, is that our customers have to be right. For example, we put a lot of effort into making sure that UpToDate Expert AI is making the right guidelines and recommendations. And the key to doing that is to make sure we ground the models.

We use trusted and verified content from UpToDate, and we do not compromise on the ability to audit, understand, and explain. Where did these guidelines and recommendations come from? Our customers can always trace back to the source, and that has been key to our thinking.

Trusted AI when you have to be right

I go back to Wolters Kluwer’s key value proposition of When you have to be right. We are really taking that principles-first, disciplined approach. Yes, you want to be fast to market. Yes, you want to be competitive, but you cannot compromise on privacy, security, and the accuracy of these algorithms.

You still have to take that same very focused and disciplined approach, understanding that the technology is moving very fast, and you have to keep up - but you can't compromise on those key tenants. You have to be right, the content has to be trusted, verified. Trusted AI is key.

AI that can be trusted

We worked very closely with our customers to make sure that they understood how the technology worked, how we were going to deploy it, how we would maintain that high standard, how we'd make sure that the technology was able to be validated and we could maintain that human expert in the loop when it was necessary.

I think that really helped overcome some of the concern around AI, and also brought in that obvious value statement so we could see what the technology could do and how it can improve things.

We also took a disciplined approach, moving from what might be considered the easier, repeatable, automated tasks that generate a lot of toil and then slowly moving up the value chain to more complex scenarios like decision support that would be really key to have that expert in the loop.

AI fifteen years ago… 

Wolters Kluwer created an AI Center of Excellence over 15 years ago. There was a need. If you think back to that time, it was the dawn of the Big Data era, and we recognized early on that there was a lot of information overload.

There was a challenge for professionals to keep up with this volume of information. We saw AI as a really important opportunity to help, assist, and build new solutions to deal with this deluge of information.

AI with the expert in the loop

You really have to look at the use case. What is the setting, what is the meaning and the intent of the solution, and where does the risk come in? And you have to be very transparent, with a human expert in the loop.

We’re making sure that we're not replacing, but rather augmenting our customers, who generally are experts in their fields. We’re giving them the ability to verify themselves, give them the proper context, and make sure the machine is not taking over. That is really key. It helps mitigate the risk and helps improve the experience and the solutions over time.