Smarter and conscious use of data thanks to generative AI

When ChatGPT became widely available at the end of 2022, the world was flooded with enthusiasm and skepticism. The rise of generative AI also led to questions in education. Is this a threat to critical thinking and analysis skills? Or an opportunity to organize education smarter and more effectively? Data science expert dr. Koo Rijpkema sees the latter in particular. Provided we approach it wisely.

With his more than 30 years of experience as a teacher and researcher at Eindhoven University of Technology and years of dedication as a course leader of the data analysis courses at PAOTM, Koo helps technical professionals to use and analyze data in the right way. The world and therefore also data science, based on traditional AI such as machine learning and pattern recognition, has gained a skill since the arrival of generative AI: the ability to independently create text, code and other output. We spoke to Koo about his vision on generative AI and its application in data analysis.

From hype to practice
Koo immediately saw the potential of generative AI in 2022. "When ChatGPT was launched, I thought: what can I do with this in my education?" Generative AI has now been integrated into all his courses, as a tool within the existing curriculum. He works with examples and prompts that participants can copy and try out themselves. "Suppose you have a dataset with a hundred measurements and you want to know whether there is a significant difference between two groups. Then you can ask Copilot, for example: which method is suitable, and how do I implement it in Minitab or Python?" Getting the answer is not the goal, Koo emphasizes. The point is that participants learn how to ask a smart question and then critically assess the answer. His courses are not trainings in ChatGPT or Copilot, but data analysis training courses that focus on the responsible use of generative AI.

Smart questions
Koo makes an important point here: the risks of blindly trusting generative AI. "I still see too often that people think: I'll throw a question into ChatGPT and that's it. But if you don't know what you're asking, you won't get a usable answer. And worse: if you can't assess the answer, you're at the mercy of what is essentially a text predictive machine.“ That is why he teaches students to consciously deal with prompts, to consider alternative answers, and above all: to learn to understand how a technique for data analysis or predictive modeling, for example, works. "Not every detail, but the foundation. Only then can you take responsibility for what you deliver", says Koo.

From beginners to senior professionals
Koo sees clear differences between the participants in his lessons. Young participants often have a low threshold for technology: they quickly experiment with it. In business environments, this is different. "There you see that people are happy with concrete examples, prompts that they can save and reuse later. I include that need in my teaching materials." He sees in both contexts that generative AI can accelerate the learning process. Where participants used to spend a lot of time on scripting, they can now come up with an initial design more quickly with the help of Copilot. "That means you have more time for the real content: interpreting the results, drawing conclusions.”

AI is not a black box
Despite his enthusiasm, Koo warns against complacency. "AI models can hallucinate: generate convincing-sounding nonsense. And if you accept and present it as true, then you are and remain responsible for it, not the system! Generative AI is really a co-pilot, an intelligent assistant that helps you navigate, but you remain the one holding the steering wheel." An attentive attitude also means: do not upload sensitive data to public tools, always assess yourself what you enter and extract, and consider ethical implications.

Learning together
In addition to his open courses, Koo also includes the subject in his in-company training courses, tailored to the group he is working with. He deliberately brings together participants from different departments. "This not only results in substantive enrichment – for example, if someone from process chemistry works together with someone from quality control – but also sustainable internal networks.“ During a follow-up meeting, in which participants show what they have done in practice with the new insights, you really see the effect!

A role model with a critical eye
Despite all the possibilities of generative AI, Koo believes that enthusiasm should be tempered a bit. "Not to block it – certainly not – but to use it consciously."" He refers to the book Human Compatible by AI guru Stuart Russell, who warns against a society that surrenders itself too quickly to machines. "Everyone who works with AI should read that book," he says. "It is clear, critical and written by someone who really knows what he is talking about.”

Stay curious and stay human
What is Koo’s most important message? "Don’t get addicted to generative AI. Use it wisely, as a tool. Make sure that you always know what you are doing, why you are doing it, and what its impact is." Generative AI is not only a technological breakthrough, but also an opportunity to make people better, smarter and more aware. As long as we remain the pilot ourselves!

The newest knowledge and expertise
Curious about Koo’s courses? PAOTM offers various courses in the field of data science for both beginners and advanced participants. During your participation, you will also learn how to use generative AI responsibly in your daily work and thus get the most out of your data.
View the program below. Does a course not have a date? Register your interest! Or let us know when your team has a knowledge question; we are happy to help you with a tailor-made in-company course.

Book tip:
Stuart Russell – Human Compatible: Artificial Intelligence and the Problem of Control (Penguin Books). An accessible and sharp book about the ethical side of AI, written by one of the founders of the field.