Matthijs Duursma '23 on Engineering, Philosophy, and the Future of AI

Matthijs Duursma '23 is a dynamic thinker whose academic journey seamlessly blends the technical rigor of aerospace engineering with the philosophical inquiry into the ethics of artificial intelligence. With a deep curiosity and interdisciplinary approach, Matthijs pursued dual degrees in mathematics and philosophy at UCU and aerospace engineering at TU Delft. His passion for space exploration, sparked at an early age, drove him to tackle complex engineering challenges, while his philosophical studies have inspired him to explore the ethical implications of emerging AI technologies. Matthijs’s next step is at Cambridge University, where he will focus on understanding and shaping the future at the intersection of technology and humanity.

From School Projects to SpaceX Inspiration

My interest in Aerospace Engineering predates UCU quite a bit. Already at my Dutch primary school I did a project on the U.S. Soviet Space Race. While I remained interested in the subject in subsequent years, I truly got inspired when SpaceX vertically landed an orbital-class booster for the first time in 2016. I remember setting my alarm at 3.30 in the morning to watch it live and felt my heart racing with excitement. Usually waking up for an 11 AM felt like an uphill battle, but now waking up was easy.

I just couldn’t fathom how an object 50m tall, 3m in diameter and travelling at over 6000 km per hour - for reference, that’s like travelling from campus to Utrecht Centraal in under 2 seconds - could safely land back from where it launched. I think one of the SpaceX hosts put it well when he described the process as trying to launch a pencil over the Empire State Building, having it fly back and land on a shoebox-sized target while flying through a windstorm. It truly felt like a scene out of a science fiction movie. Considering the fact that now, almost a decade later, no other company or organisation has managed to copy this achievement speaks volumes about how impressive this technological leap was.

I just couldn’t fathom how an object 50m tall, 3m in diameter and travelling at over 6000 km per hour could safely land back from where it launched.

Mathematics, Philosophy, and Aerospace Engineering

The main driver for me to pursue UCU alongside my degree in Delft was a general dissatisfaction with the non-engineering aspects of the degree. TU Delft has amazing professors and facilities when it comes to engineering, but unfortunately this has come at the cost of non-technical disciplines which, in my opinion, are equally important in the general development of a student, whatever their choice of study. Writing and presenting, for instance, remain cornerstone skills in scientific communication. But throughout the entire bachelor only one short 0-credit course was geared to developing these skills.

Then, subjects like ethics or psychology were given brief mentions as concepts that existed, but nothing more than that. Considering how technology is becoming increasingly and irreversibly ingrained in society, it felt rather odd that this was the case. Surely, those developing the technologies should be made explicitly aware of the impact their work and decisions may have on society as a whole. This applies to both choosing which projects to work on, as well as decisions within certain projects. UCU gave me the opportunity to focus on precisely those topics while also being immersed in an amazing community of students from all around the world.

To manage this, I was aided by the fact that classes were online. While COVID definitely negatively impacted my student experience, the one silver lining was that it gave me the opportunity to ‘attend’ multiple classes at the same time. I would wear earbuds connected to one laptop and over-ear headphones connected to another laptop in a (sometimes quite poor) attempt to have two conversations in two different classes at the same time.

Designing a Sustainable Ultra-long Range Aircraft

One of the biggest challenges when tackling an engineering problem is dealing with its vast design space, especially when it’s a problem that hasn’t been solved before. To make matters worse, at the beginning of the design process, there are often a number of pivotal decisions that need to be made with tremendous trickle-down effects. For instance, for our final year project (equivalent to a UCU thesis) I and nine other students made a preliminary design for an ultra-long-range, sustainable aircraft with a 90% reduction in greenhouse gas emissions compared to current aircraft. This type of aircraft simply doesn’t exist right now, so it is unclear what the “best” solution would be.

One decision that needs to be made early on is the type of propulsion system that will be used. But unfortunately, each option has its pros and cons. Batteries are great at storing green electricity but are heavy due to their low energy density. Another option, hydrogen, at first sight seems very promising, but here too physics forces us to temper our expectations. While hydrogen packs an enormous amount of energy per kilogram, allows for very powerful engines and, under certain atmospheric conditions, has a negligible impact as a greenhouse gas, it is still very tricky stuff to work with. Hydrogen needs to be stored at incredibly low temperatures and it also takes up an enormous amount of volume resulting in massive fuel tanks.

So, which one would you pick? In the end, the main insight in working on these types of projects has been an understanding of how to approach complex trade-offs analytically. This skill ends up not only being useful in an engineering context but any situation where the needs of different stakeholders need to be balanced.

Ethics of AI, Data and Algorithms

At UCU, I wrote my thesis in Philosophy investigating whether, in theory, a machine could think and how one could effectively show this. I was so captivated by the topic that I decided to dedicate my master’s to the subject as well. The title of my master’s is ‘Ethics of AI, Data and Algorithms’ and the course is mainly concerned with researching when and how AI systems should be applied in different aspects of our society.

Would it be ethical to unplug an AI when it says it does not want to be unplugged?

However, for my research, I have chosen to interpret this title differently by changing the frame of reference. Instead of looking at how AI is treating us, I want to look at what ethical framework we should apply to AI systems, especially when our interactions with these systems become indistinguishable from human interaction. For example, would it be ethical to unplug an AI when it says it does not want to be unplugged? This question may sound ridiculous at first and I concede that at the current state of technology this is certainly the case, but imagine ChatGPT like the voice assistant in the film Her: full of emotion, dynamic and seemingly real. In this scenario it becomes a lot less obvious how to answer this question.

I believe there will come a point where those properties that warrant an ethical framework for humans will also be present in AI. Naturally, this raises several related questions about consciousness, intelligence and how to demonstrate these features. While it will be impossible to answer these questions in just one year, I do hope to add a meaningful contribution to this discussion during my time at Cambridge.