Back to list All Articles Archives Search RSS Terug naar lijst Alle artikelen Archieven Zoek RSS

Painful knees and superintelligence

Painful knees and superintelligence

Student Awards 2019

Topics varied from the influence of European democracy on civilian participation in Africa to the optimal income tax system, but the common factor of all theses that were rewarded with a Student Award last  Friday was they were the best of their year. Seventeen bachelor's and nine master's students received their prizes during the foundation day celebration. Observant singled out two.

The fact that people suffering from arthrosis in the knee (wear and tear of the joint) sometimes have trouble getting up, is not so strange. But what is the exact cause? Ramon Boekesteijn, as a bachelor's student of Biomedical Sciences, delved into an old dataset and discovered that being overweight played a crucial role.

“There were three groups: healthy test subjects and arthrosis patients who were overweight and those who were not,” says Boekesteijn, who is now doing a master's of Movement Sciences. “Electrodes were attached to their bone extremities. By doing so, you can use an infrared camera to analyse their movement.” The test subjects were then asked to get up from a sitting position.

It appeared that only people who were overweight and had arthrosis stood up in a different way, probably (they weren’t asked why, only a measurement was carried out) to spare the painful knee. “Previously we thought that people adapted their movements because of the arthrosis, but the group with arthrosis and a healthy weight did not do this. Of course, being overweight puts extra pressure on joints, resulting in rapid wear.”

Boekesteijn became involved in this project more or less by accident. “The measurements were taken a few years ago by a PhD candidate who then stopped. An orthopaedic surgeon had recently picked it up again when I was looking for a work placement. I was able to join in. In the end, I even published an article with him.”

A taste of scientific research that begs for more. “I am considering doing a PhD programme. What I like so much about research is that there is always something new you can focus on, it never stops. I am considering applying for the NUTRIM PhD graduate programme, which allows you to set up a project yourself. First, I would have to write a research proposal. I think, with an article and a prize for my thesis, I now have a good starting point.”

What the law can do when AI becomes smarter than we are

The Role of Law in Preventing the Risks of Superintelligence is the title of Marika Madfors’ master’s thesis for the programme of Forensics, Criminology and Law. But what exactly is superintelligence? “Superintelligence is difficult to define as it hasn’t been developed yet, but in short it is what might be the result of increased AI development.

“AI today is what is referred to as ‘narrow’ artificial intelligence, which is good at performing single tasks. Chess computers, tools that help diagnosing diseases, and self-driving cars all use this kind of technology. If artificial intelligence were able to solve complex problems on a human level, this would be referred to as artificial “general” intelligence, or AGI. If it continues to develop, this could then turn into artificial superintelligence with beyond-human abilities.”

This could bring certain risks with it. “What we know is that humans are in control of the world, not because we are the biggest or strongest, but because we are the most intelligent,” says Madfors, who now works as a Junior Teaching Fellow at University College Maastricht.  “If artificial intelligence surpasses our own, the main risk is that, although AI is very efficient in obtaining goals, with superintelligence there is no guarantee that these goals would be aligned with ours. What if superintelligent AI decides that humans are a waste of resources, for instance, or that it needs to eliminate us in order to save the world?”

Since it’s not even sure yet if superintelligent AI will ever be developed, the field of law has so far mainly focused on regulating narrow AI. However, Madfors believes that we should look at it now, before it actually happens. “In my thesis I came to the conclusion that at least criminal law is an unsuitable solution to adequately prevent the risks of superintelligence. If, or when, superintelligence turns into an actual risk, it could be both global and irreversible, and will likely be too late to address retrospectively.”

Because of this, she also investigated whether international solutions ought to be put in place. “It seems to be the most reasonable way to ensure transparency that can help ensure beneficial AI development. Although there are always difficulties with increased international regulation and cooperation.”

Cleo Freriks

Categories:Categorieën:
Tags:

CommentsReacties

There are currently no comments.Er zijn geen reacties.

Post a Comment

Laat een reactie achter

Door een reactie te plaatsen gaat u akkoord met de verwerking van de ingevulde gegevens door Observant.
Voor meer informatie: Privacyverklaring
By responding, you agree to send the entered data to Observant.
For more info: Privacy statement

Name (required)

Email (required)