Prof. dr. Rabaey: ‘Get rid of traditional, digital and deterministic thinking’

ENGINEERINGNET.BE – Professor Rabaey (photo) is invariably focused on the future. This is how it has been for decades: He has been called a visionary more than once.

In the mid-1990s, for example, he developed a predecessor to the iPad: the Infopad. Today, he focuses on the convergence and fusion of humans and technology in, as he puts it, the generation after the next computer.

The demand for computers is growing exponentially, and that speed will certainly not change in the coming decades. But it also means that the need for energy will increase sharply, according to Rabaey. What’s more: The energy to be used for our computer use will make up a very large percentage of the total energy consumption of our society.

That makes the situation unsustainable. There are then two possible solutions: either you stop the number of computer solutions and let our computer use become saturated, or we must look for more efficient energy solutions.

That concern is not new. With each new innovation, we take a step forward in terms of energy consumption, says Rabaey. “But there are limits to how far you can go with it. And once that limit has been reached, the question naturally arises: what now? ‘

Our biological computer
According to him, it is high time to look at computing in a different way. “There are several paths we can follow. But a very interesting track is nature and biology. Our brain is a very efficient machine. Your brain’s energy consumption averages 20 watts. Compare that to classic bulbs: you could not get much light out with such a watt. ‘

Although our brains have a limited power budget, our brains are raining a lot. Think of it as our biological computer. ‘On the one hand, it’s comparing apples to oranges, I’m aware of that. On the other hand, there is a huge opportunity in it: we can look at what is happening in our brains for the technological solutions of the future and learn from it. ‘

Take the neurons in our brain for example. These cells in the nervous system are the building blocks of the brain. Neurons can transmit impulses to each other through synapses. But sending electrical impulses is very expensive from an energy perspective. Our brains are therefore also dependent on chemical operations: they combine electrical and chemical data processing. ‘

Another striking example: ‘In our brain, the processor and memory are not divided, unlike computers today. It also ensures that our biological computer is much more efficient. ‘

Local computing
So, of course, the question arises, is actually connecting technology with biology a far-from-our-bed show? Not exactly, according to Rabaey: There are already plenty of experiments with the aforementioned principles.

‘In terms of computing, there are different domains. On the one hand, you have Google and Amazon in this world, which are mainly dependent on cloud processing. Any task sends a signal to the cloud. They require a huge amount of power – gigawatt power – because they are organized to perform as many tasks as quickly as possible. Despite the sky-high energy costs, the classic computer works best there. ‘

At the other end of the spectrum you will find data processing as in your smartwatch. To take full advantage of the features of the smartwatch, it is already important that we adapt the technology in such a way that it fits in nicely with our body, says Rabaey.

‘It is not speed and size that are the key, but to perform the right task at the right time, by receiving the right signals. The closer you get to nature, the more local data processing becomes more important, and the more the operation of computers will resemble and interfere with our biological computer. ‘

Smart prostheses
We will therefore see organic data processing for the first time in IoT in our home environment, such as automatic temperature control or your lighting. But also in wearables and implantables, such as smart earpods.

Another challenge is to make dentures even more intelligent. In people with an amputated leg or arm, EMG signals can already be measured from the muscles left after the amputation. These signals are then used to control the prosthesis’, according to Rabaey. “But it requires the prosthesis user to concentrate a lot on the activity, which makes the process extremely tiring. Most users of these prostheses stop using them after a while: they are stressed and exhausted. ‘

By using organic data processing, scientists were able to optimize this process. ‘By connecting a chip to a network of neurons, for example, you create brain-machine interfaces. The next generation of prostheses is about automatic behavior: If you opt for a super-local, very efficient and very closely connected – possibly implemented – processor, the prosthesis will really be an extension of the body. ‘

Deterministic digital thinking
When you start up a computer today, you expect the machine to perform requested tasks in exactly the same way every time, thus providing the same answers every time. Rabaey: ‘Deterministic thinking is the easiest way. Because since the entire test infrastructure is also built around this, you can now renew yourself within a nicely defined framework. ‘

He ends with visionary words with a call to color outside the lines. “In the future, we must let go of traditional digital thinking and dare to build computers that are stochastic and can reason themselves. We are not deterministic as human beings. (smile) It requires a total shift in both the operating model and the business model, but that effort will lead to innovative, sustainable solutions: Sometimes you have to dare to jump in to push a society forward. ‘

Leave a Comment