“Energy consumption places limits on digital computing”
Wilfred G. van der Wiel, Professor of Nanoelectronics and Director of the Centre for Brain-Inspired Nano Systems (BRAINS) at the University of Twente, is considered to be a pioneer in the field of Material Learning. He has a second professorship at the Institute of Physics at the University of Münster. The focus of his research is on unconventional electronics for efficient data processing. He spoke to Christina Hoppenbrock about the potential of intelligent matter and the future of digital computing.
There is no standard definition of intelligence. In general, we can say that it is the ability to process and remember information and use it later in a different context. In connection with matter, it is a minimalistic definition which has nothing to do with emotions or consciousness. To illustrate what intelligent matter is, we can divide materials into four levels. Level One means that the functionality is dependent on the structure of the material. One example is spectacles with precisely ground glasses. Level Two is what we call “responsive”, when the material responds to certain environmental stimuli, for example glasses which darken in the sunlight. Level Three is when a material processes several environmental stimuli at the same time and has a different response depending on the stimulus constellation. A hypothetical example would be spectacles which darken to an extent depending on the amount of sunlight and which simultaneously adapt their colour to the colour of the wearer’s clothing. It is only in connection with Level Four that we talk about intelligence – when the material “learns” through experience. In addition to the properties mentioned, such spectacles would also have to be able to recognise what the wearer’s needs are at that moment: whether for example he or she is driving car and looking into the distance or is reading a book. The spectacles could then adjust the visual acuity correspondingly.
Your particular interest is in innovative computer hardware. What does this have to do with intelligent matter?
Traditional computer architecture is entirely unsuitable for certain computations, for example for artificial intelligence (AI). The reason is that in AI an extremely large amount of stored data has to be transmitted to the processor and stored again – for example if a computer is trained to recognise people on images. If storage and processor are separated spatially, that costs a great deal of time and energy. What’s much more efficient is a network which works in a similar way to nerve cells in the brain, storing and processing information in one place. We want to develop new types of computer components using intelligent matter. That will be a long process, but the brain is proof that there are other solutions. Our computer chips work with networks of nanoparticles whose behaviour is specifically influenced by molecules. Such chips will not only permit faster computations but will also have much better energy efficiency.
Your aim is to overcome the limits of digital computing …
Right. Currently, these limits are defined exactly – by energy consumption. For example: real autonomous driving in everyday life is not very attractive with today’s computer technology because most of the energy in the battery is used up for computing purposes.
Intelligent matter is still a long way off. What will it be able to do in the future?
It’s difficult to make predictions. Our solutions may only take hold in perhaps ten or fifteen years. But the idea is already there that we should take the brain as our model, and we see the trend today in various electronic devices. I’m convinced that this development will catch on. However, we don’t want to replace everything. Digital technology is very good today, especially when precision is called for. What we want to do is improve certain functions and make them more efficient.
This article was first published in the University newspaper “wissen|leben” No. 4, 1 June 2022
Collaborative Research Centre (CRC) 1459: Intelligent Matter
Producing intelligent matter requires interdisciplinary collaboration involving specialists from the fields of Physics, Chemistry, Material Sciences. Biology and Computer Science. 26 working groups from the University of Münster, the Max Planck Institute of Molecular Biomedicine, and the University of Twente are all collaborating in CRC 1459 to make this project a reality. In their work they examine various classes of materials: molecules, soft materials such as polymers, and solids. Their aim is to make matter intelligent.
Contact: Dr. Christina Kriegel (Managing Director), phone: 0251/83-34521, email: ckriegel@uni-muenster.de
Save the date
The “First Münster Symposium on Intelligent Matter” will be held on 22 June, starting at 9:50 a.m., at the Center for Soft Nanoscience (SoN) at Münster University. International experts will be presenting the results of their work in this field. There will also be a poster session in which members of the CRC will be talking about the progress made in their projects.