Key takeaways
- According to the Dutch Scientific Council for Government Policy (WRR, 2021), AI is a system technology, comparable to the steam engine, electricity, the combustion engine and the computer.
- A system technology is present everywhere, becomes ever cheaper and offers a platform for unexpected innovations.
- Using data to improve the existing business is commonplace; too few organisations use data to develop new products, services and business models.
- Feldberg believes in hybrid models: the algorithm supports the decision, the human ultimately decides.
- Computers cannot think: they calculate and recognise patterns better than people, but lack imagination and empathy. Legal and ethical frameworks are needed for responsible use.
Never underestimate the power of data-driven innovation
“Organisations massively underestimate the power of data-driven innovation,” says Frans Feldberg, professor of Data-Driven Innovation at Vrije Universiteit Amsterdam. Using data and analytics to improve the existing business is commonplace, but too few organisations seize the opportunity to really innovate by developing new data-driven products and services. “You no longer have to explain to managers how important data is. Most organisations use data to optimise the supply chain, for example. That you can also use data to stay relevant by developing new business models is a real eye-opener for many.”
What is a system technology?
Feldberg: “Moore’s law (a doubling of the number of transistors on a chip every two years) still applies. And according to the WRR, artificial intelligence is not just a technology. It is a system technology, best compared with the rise of the steam engine, electricity, the combustion engine and the computer. These lead to unexpected applications. Edison could never have predicted that electricity would eventually be used for brushing teeth. AI is in your phone, car, washing machine, the road surface and lamp posts. A system technology offers a platform, a fertile breeding ground for all kinds of new innovations.”
A characteristic of a system technology is that it is present everywhere and becomes ever cheaper.
AI in healthcare: the algorithm sees more
A concrete example: “According to scientifically validated benchmarks, algorithms already beat humans in detecting suspicious tissue that is not yet visible to the human eye. That is not strange, because an X-ray has some 4,000 shades of grey, which the human eye can never all distinguish. When it comes to recognising patterns in data, the computer beats the human.” Developments go further still: personalised medicine, with treatments tailored almost at cell level, is only possible thanks to AI and increased computing power.
Another example is the intensive care unit. If you discharge someone too early, the risk of readmission or mortality rises; an unnecessarily long stay costs money and bed capacity. There are algorithms that support intensivists in deciding the best moment of discharge; intensivist Paul Elbers of Amsterdam UMC is researching this.
Hybrid models: the algorithm supports, the human decides
This does not mean that hospitals rely blindly on the algorithm. Feldberg believes in hybrid models, in which algorithms support people in taking good decisions. On transfer or discharge the doctors first decide themselves. Then they look at what the algorithm says. If the advice differs, they meet again. The algorithm is fed with all available data, such as blood pressure and heart rate measurements and the reports of nurses and doctors, and thanks to reinforcement learning it becomes ever smarter. “People play a crucial role in decision-making; you do not want to leave bad-news conversations to a computer either.”
Feldberg also has an eye for the dark sides of AI. He argues for responsible AI and critical reflection on the role of big tech. The ease with which people give away their data to companies such as Facebook and TikTok surprises him. “If the product is free, you are the product.”
What the algorithm does versus what the human does
The division of roles in a hybrid model.
The algorithm
The human
Strength
Calculating and recognising patterns
Imagination and empathy
Role in the decision
Supports with a prediction
Weighs up and decides
Input
All available data
Context and experience
Responsibility
None
Remains with the human
The great prediction machines
Inspired by the book Prediction Machines by Ajay Agrawal and colleagues, Feldberg sees a large role for predictive algorithms. Predicting is filling in information you do not have on the basis of information you do have. Think of a jigsaw: the further you get, the fewer pieces are missing and the easier it becomes. ChatGPT works this way too: it is one big prediction machine looking for the right answer.
He regards the algorithm registers that Dutch municipalities, provinces and other public bodies have to keep as a logical and good development. Legislation is also coming from Brussels: the Digital Services Act, the Digital Markets Act and the Data Act. A kind of quality mark for algorithms could help, but there are major challenges. “Some algorithms are black boxes, and others are really understood by only a limited number of people in the world. As a society we have to think carefully about what we want with AI and what the guiding principles should be.”
Data-driven innovation in energy and retail
In the energy sector Feldberg sees micro grids and smart grids in which smart innovations and data play the leading role. People set up energy cooperatives. At macro level supply and demand are matched with AI to prevent congestion on the grid, and algorithms indicate when preventive maintenance is desirable.
The power of data-driven innovation is clearly visible in retail, where subscription-based services are emerging in which large amounts of customer and usage data are collected, whether or not through the internet of things. At Coolblue you can take white goods in a full-service model for a fixed amount per month. In the car industry options are increasingly sold as a subscription: BMW charged 19 euros a month for seat heating. Stellantis, the parent company of Peugeot, Citroën and Opel, expects software-enabled services to generate around 20 billion euros in additional annual revenue by 2030.
ChatGPT did not come out of the blue
The effect of Moore’s law is clearly visible in ChatGPT. We now have the computing power to analyse gigantic amounts of data, discover patterns in them and derive knowledge. OpenAI was founded in 2015. ChatGPT, too, is essentially one very large prediction machine. That requires enormous computing power, with high energy consumption and extra CO2 emissions. Google’s business model is under pressure: “If you can choose, which would you rather have? A fully written, to-the-point answer to your question, or a list of ten search results of which the first five are sponsored?”
Data-driven innovation does not mean you have to reinvent everything. Proven concepts such as ideation and design thinking still apply. When Pon, winner of the Dutch BI & Data Science Award, introduced the start-up Lensor for automatic damage detection, a brainstorm produced the idea of applying the artificial intelligence not only to Pon’s own cars but also to company vehicles and rental cars of external parties.
The essence of “resourcing with data”
At the KIN Center for Digital Innovation of VU Amsterdam, Feldberg and colleagues research data-driven business model innovation. Wendy Günther and colleagues examined how organisations create data-driven value propositions by reusing data in a different context, described in the award-winning article Resourcing with data: Unpacking the process of creating data-driven value propositions.
Computers cannot think
Data-driven innovation is a necessary condition for organisations to stay relevant. Yet Feldberg adds an emphatic caveat. “Computers cannot think. They have no imagination or empathy. If we think that computers can think, and start using them to think for us because we think they can, things go wrong!” Computers, algorithms and chatbots can perform certain cognitive processes, such as calculating and recognising patterns, much better than people. But that is something quite different from thinking. That is why legal and ethical frameworks are needed to safeguard the responsible use of AI.
How Diamond Decisions applies the hybrid model
The decision stays with people. In Business Modeling it is determined which decisions an algorithm may support, data analytics provides the prediction and its substantiation, and in Decision makers people learn to weigh that advice and take the decision themselves.
Talk to the author
Daan van BeekDiamond Decisions
Discuss your question with Daan
Tell us about the choices you face. In a no-obligation conversation we discover together which decisions make the difference in your organization.
Schedule an introduction →Step 1 of our approach
Diamond Inspirations
The spark for decisions with impact: in an inspiration session you discover which decisions make the difference in your organization.
Explore Inspirations →Frequently asked questions
01
What is a system technology?
A system technology is a technology that penetrates everywhere, becomes ever cheaper and forms a platform for many other innovations, such as the steam engine, electricity and the computer. The WRR classifies AI as one.
02
What is data-driven innovation?
Data-driven innovation is using data not only to improve existing processes, but to develop new products, services and business models, such as subscriptions based on usage data.
03
What is the difference between a hybrid model and fully automated decision-making?
In a hybrid model the algorithm gives advice and the human decides and remains responsible. In fully automated decision-making the algorithm decides itself, which is only appropriate for simple, low-risk decisions.
04
What does this mean for a small or medium-sized business?
You do not have to develop AI yourself: it is increasingly built into the software you already use. The question is for which decisions you use it and whether your data is good enough to rely on the outcome.
Further reading
Sources
- WRR (2021): Opgave AI. De nieuwe systeemtechnologie (Mission AI: The New System Technology).
- Agrawal, A., Gans, J. & Goldfarb, A. (2018): Prediction Machines: The Simple Economics of Artificial Intelligence. Harvard Business Review Press.
- Stellantis (2021): around €20 billion in incremental annual revenues by 2030 from software-enabled vehicles.
- Günther, W.A. et al. (2022): Resourcing with data. Journal of Strategic Information Systems.
About the author
About Diamond Decisions
Diamond Decisions helps organizations become better by enabling better decisions. We don’t start with data, dashboards or AI, but with the decision that needs to improve, and work back from there to KPIs, data and technology. We do that in four steps: inspiration and alignment, business modeling, analytics and data platform and data-driven decision makers. That way BI and AI don’t remain an IT project, but deliver value every day.


