Sjoerd van den Heuvel (Analytics and AI)

From Excel to AI, but how?

"What do you think of when you think of AI? And what is already happening at work?" Sjoerd van den Heuvel gets straight to the point with his presentation during the Verbondscourse. In any case, he is crystal clear about his intentions. "Go on a journey with AI, but do it in a structured way."

Van den Heuvel does not turn his heart into a murder pit. "I have seen so many organisations fail in the last fifteen years. And do you know why? Because AI is usually not embedded in the organisation. Be honest: how many of your leaders already have a concrete picture of AI and a concrete ambition in this area? That dot on the horizon is just so crucial. You really have to work the right culture, manage expectations and ensure clear communication. Because you encounter the necessary emotions along the way. One colleague fears for his job, while another thinks AI is very cool."

Covenant Course 2025
Van den Heuvel stands in front of a full house in Résidence Groot Heideborg in Garderen. He takes the 28 participants in the Covenant Course 2025 on an interactive journey of discovery. Everything shows that as an associate professor of Data-Driven Business & People Analytics , he is used to being in front of the class. He talks easily, quickly and flies through his presentation. Like a real teacher, he had also given today's students homework. Among other things, they had to answer the following three questions. What are examples of AI applications that you have implemented? In what process? And what has it yielded?

"You have to work very the right culture, because you encounter the necessary emotions. One colleague fears for his job, while another thinks AI is very cool"

Learning and experiencing
The answers vary enormously. Some are still at the beginning and are individually working with ChatGPT or CoPilot, but hardly any steps have been taken in the organisation. Others have started cautiously and limit themselves mainly to tools such as speech to text for recording conversation reports, summarising and classifying. And still others use AI frequently, including in customer contact and in claim assessment. The majority, about three-quarters, are working on a pilot internally, while a few have already completed the necessary use cases . Van den Heuvel emphasises that 'just do it' is the most important thing. "Of course, there are also dark sides to AI, but innovation will not get off the ground if you put up all kinds of obstacles right from the start. AI has to be learned and experienced. And you have to try to enjoy it."

How many FTEs?
It's time for another question. Van den Heuvel: "Suppose that all the effort you put into creating, editing and discussing graphs and tables in your organisation adds up. How much time and therefore money (FTE) do you spend per year?"
Mutual consultation follows. Ten percent say one. I'm guessing at 25 percent, replies another. "Percentages don't hurt," Van Den Heuvel responds wittily. "How many euros are you talking about?" 25 percent amounts to about seventy people. And the ten percent relates to one tenth of a hundred, so ten FTEs. "Just calculate what it saves an organisation per year if AI does that for you. There are simple dashboarding and analytics solutions, such as Visier and Crunchr. They cost you a ton or two a year, but then you can make a big saving with them."

"Just calculate what it saves your organisation per year if AI does a lot of work for you"

Ethical aspect
"Who does consider himself ethically responsible?" is Van den Heuvel's next question. A few fingers go up in the air and he points to one of the male participants. "Do you also think you can drive well? You get to drive a self-driving car, which we equip with your ethical awareness and your driving skills. You are given one assignment: you have to collide. In the middle is a tree, on the right an old grandfather and on the left a 6-year-old girl. What do you choose?" He chooses, of course, the tree.

Self-driving car
"I had forgotten something," says Van den Heuvel, "because you are not alone in the car. The whole family is there. What do you want? Still the tree?" Yes, still the tree. "The whole family dies in that collision. So what do you want? Grandpa or the child?" It will be grandpa.
"This is a real and moral issue," Van den Heuvel emphasises, "with which we determine the value of life." He shows a list, ranging from girl, boy, homeless person, dog and doctor. "This list comes from a study known as The Moral Machine Experiment and you can see it for yourself: 1 is the pram, 2 a girl, 3 a boy and 4 a pregnant woman. Fairly at the bottom, just above the cat, the criminal and the dog, we find 'an old man' and 'an old woman'. You get the point that comes around the corner with self-driving cars? Who will be liable for the choices that the car makes?"

Summarise
Time for the next issue: recording, transcribing and summarising. "Who is already doing this?" About five or six hands go up in the air. One of the ladies present asks if and how emotion plays a role. "Does AI recognise that? For example, if someone says very sarcastically: I am really very happy with your service!" According to Van den Heuvel, AI can already recognise emotion, especially if video analysis is used. "But," he emphasises, "of course sometimes things go wrong."
An example. In Finnish, hän means both he and she. If you let Google translate a number of sentences, that translation is also determined by the bias that we ourselves have put into the data on the internet. And how. This is the result. He is the boss. She is naked. He is an astronaut. She is a cleaner. He works. She takes care of the children. "Shocking," says Van den Heuvel, "but unfortunately still the reality."

"If you let Google translate a number of sentences, that translation is also determined by the bias that we ourselves have put into the data on the internet"

Translation
The example of translation comes up more often during Van den Heuvel's presentation. Rightly so, especially when you look at the increasingly international context in which many organisations operate. He shows the Course participants a video of himself, in which he passes by in all kinds of languages. "How cool is that for an international organisation. You make a presentation once, in English, and can then address everyone in his/her own language. Also consider a training video, which you only have to record once from now on. Or an announcement by a CEO in an international company. It's all possible and easier than you think."

Google Glass
Perhaps the most impressive video shows Van den Heuvel at the end. A deaf girl puts a Google Glass on it. This converts the spoken language into written language, so that the girl can suddenly communicate directly. "The world in the customer and employee field is at your feet with these kinds of tools," says Van den Heuvel. "The Google Glass may cost three hundred euros, but it ensures that many people who cannot keep up now, for example because they do not speak the language, can suddenly keep up. Why shouldn't an insurer open up its organisation to people with a distance to the labour market. Buy a hundred of those glasses and make sure you have positive branding."

"The question will soon no longer be: How many people do you employ? but How many Agents work for you?"

AI Agents
However, the hype of this year is/will be AI Agents. "Really a keeper," said Van den Heuvel. "It's actually a digital employee that combines the best of automation and artificial intelligence. And thus saves a lot of time. The possibilities are endless. Think of reading CVs and then giving a recommendation of a top 3. Or, another example. The Agent goes through the medical file, which counts about 200-300 pages, of a personal injury. You will not only receive a summary, but also advice for the medical specialist. Of course, it takes a lot of work to get the Agent to do what and how you want it, but I predict that soon the question will no longer be: how many people do you employ, but how many Agents work for you?"