Paul Roos pupil Alexander van der Merwe
Paul Roos pupil Alexander van der Merwe

Young scientist off to represent SA at science and engineering fair in US

By Staff Reporter Time of article published Apr 22, 2021

Share this article:

Cape Town - A fifteen-year-old Paul Roos Gymnasium learner has been selected to represent South Africa in the 2021 Virtual Regeneron International Science and Engineering Fair (Isef) in the US.

Alexander van der Merwe will join more than 1 800 young scientists from around the world who will compete for nearly R73 million in awards, prizes, and scholarships at the Regeneron Isef from May 16 to 21.

Known as the young scientist from the Western Cape, Alexander earned his spot in the prestigious competition for his investigation into whether people can differentiate between computer-generated and human art.

The 15-year-old aspiring scientist said his investigation had been inspired by an artwork produced by a Generative Adversarial Network (GAN) computer algorithm, titled “Edmond de Belamy“, from La Famille de Belamy, which sold for R6.3 million at the time at a Christie’s art auction in New York in October 2018.

Thousands of artistic images are fed into the GAN network to “train” it to produce artistic images that are so realistic, it is difficult to tell it apart from human artwork.

“I’m interested in artificial intelligence (AI), and when I read about an artwork produced by a computer selling for $432 500 (more than R6 million) at an art auction, I thought, can it be that computers can use AI to have imagination, a characteristic that only humans should have?

“Answering this question has far-reaching implications for how we view computers’ role in society.”

Eskom Expo academic director Krishnie Naidoo said Alexander had researched a very important field that is rapidly expanding in the age of AI.

Naidoo said: “Alexander explored whether machine learning algorithms can create art that is imaginative and creative enough to fool people into believing that the art was created by a human, using a sample size of 480 participants.

“He ventured into the lesser-known field of machine learning for the creative arts. Soon we will be able to tell whether computers can learn how to reproduce human art through a set of algorithms or if creativity is exclusively a human competence; maybe even whether humans tell the difference between human art and computer-generated art.

“We will be able to learn if machines can alter our constructs of the fine arts, which are humans expressing creativity, emotion, skill and intelligence, and this also has implications for the creative arts as a multibillion dollar industry. Hence the significance of Alexander’s research goes far beyond just machine learning and art,” said Naidoo.

Cape Argus

Share this article: