The winners had almost no idea

They tried it. Just like that, with controllable prior knowledge. They weren’t even in the same class. Aargau High School students Lili Rölkötter and Emilott Suttmöller first met in computer science project classes at their high school in Wohlen — in a group where the other students already knew what they wanted to do that semester. Chess computer programming, for example. Or an artificial intelligence (AI) image generator.

Usually computer science, usually nerds, you guessed it.

But Emelotte and Lily’s journey was very different. “We had practically no idea what artificial intelligence was,” says Imlot. “And you don’t have an idea for your own project either.” But then the teacher informed us about the competition. The ETH Zurich call for young people interested in artificial intelligence was launched for the first time this year. The two students decided to participate – and immediately won the competition. Creative programming class.

That was last spring. And they still laugh when they talk about it.

Tell me what to wear

The beginner’s idea seemed to have convinced the jury: what it would be like if you no longer had to think for a long time in front of the wardrobe or wardrobe in the morning about what to wear, but instead received some suggestions instantly thanks to a smart application – perfectly in line with the weather forecast in today? This sounds impressive.

The journalist is already thinking about his wardrobe at home and a personal fashion consultant on his smartphone. Never leave the house too fat or dressed too lightly again, this will be practical! But then it becomes clear that the two worlds collide in the meeting of this text in Kanti in Volyn. The author of this article did not have computer science in school. Hence the exaggerated expectation that he will now see a ready-made application.

On the other hand, Emelotte and Lily have been coding since first grade in high school. This subject is compulsory for all middle school students in the canton. So the two students have to disappoint their conversation partner: they haven’t developed an app. That would have been too complicated. Instead, they did what computer scientists usually do: programming, compiling the code, block by block, line by line.

About K:

etc. Random_pullover, random_coat, and random_ankleboot also received their own fonts. So that the program can execute the required commands. Emmelot types her location on her tablet screen: Wohlen. And their gender: F. Then the machine starts:

Seconds later he says:

This written advice is accompanied by the gist of the project: The program displays images of pants, a shirt, and a jacket. In pixelated form, of course, but that gives you an idea of ​​how a clothing consultant at Lilly and Emmelot would one day work: He knows his users’ wardrobes because he’s saved photos of their clothes. Using this information, he could design weather-appropriate clothing, at least in theory.

Lily says: “I play a lot!”

But in practice there was a problem: the program did not recognize the photos that the two had uploaded. For example, the AI ​​mistook a shoe for a shirt. The neural network could not remember the images correctly. The students had to come up with something. Their solution: Image recognition worked well using a dataset from mail order company Zalando. The two discovered this on Kaggle, a platform for the international AI community.

This means that they were able to submit a project to the ETH competition that was at least consistent with its basic principles. And the teens agree: If it weren’t for the AI ​​competition, they wouldn’t have been able to achieve this at all. When they scored, they had a goal in mind. They were accompanied by a doctoral student from the university. But they were very independent from the beginning, says the coach. At first he just gave them some advice.

And now? Will Lily and Emelotte be part of the much-needed generation that is supposed to alleviate the shortage of technical professionals?

Lily briefly considered bringing the unfinished uniform application to the finish line. But then she had another idea: For her graduate thesis, she investigated how much aluminum in coffee capsules ends up in the coffee. Emmelot compared two literary books. “After all that programming, I wanted to do something completely different,” she says. She reads a lot and prefers to be outside rather than always glued to the screen.

Lily, on the other hand, is mostly in front of the computer. “I play a lot!” She is interested in a lot of things. So it was no surprise to anyone around her to hear that she was participating in an artificial intelligence competition. The 17-year-old has clear ideas: she wants to study at ETH, perhaps biochemistry. Emmelot has similar plans: she is interested in interdisciplinary natural sciences.

Projects can still be submitted for the 2024 AI competition. ETH announces that the timeline will have to be postponed. More information: ki-vergleich.ch.