Phenomenon around innovation

“Just as many mistakes, but people believe the AI”

People are replaced and supported: Digital linesmen are being used in tennis. In medicine, AI is increasingly being used when it comes to analyzing samples.

(Bild: Krone KREATIV/UALL/Alex Scheuber, picturedesk.com)

Advisors and assistants for homework at school, support in preparing presentations and writing texts - the use of artificial intelligence is becoming suitable for everyday use. But is the data correct? Martina Gaisch, professor at the University of Applied Sciences Upper Austria, warns against a "false" reality.

What can artificial intelligence do and what can't it do? Martina Gaisch deals intensively with this question. The head of the "Design of Digital Products" course at the Upper Austrian University of Applied Sciences campus in Hagenberg checks just as often whether what the system is spitting out is true or not.

Human qualities count
"The most important things are simply contextual knowledge and evaluation skills - these are deeply human qualities," says Gaisch. The 55-year-old sees herself in the role of refiner when she uses digitization tools to create presentations: "For me, AI is a sensational personalized learning assistant. I get creative input."

Martina Gaisch is head of the Design of Digital Products course at the University of Applied Sciences Upper Austria. She uses artificial intelligence herself, but is aware of its weaknesses.

(Bild: Dostal Harald)

While some have already recognized that the digital tools are not perfect and often hallucinate, as Gaisch calls it, there are some who trust them completely. The professor calls this group of people "unreflective" and refers to this as automation bias.

The systems are often trusted without reservation
What does this mean? Too much trust in the systems: "In the past, you had the linesman at tennis who you argued with after decisions. Today, the job is done by AI, which makes just as many mistakes, but you believe it."

Artificial intelligence is only as good as the data it has been trained with. And this training is done by humans: "AI is simply a mirror of our society, it is neither objective nor neutral."

Gaisch points out that it is predominantly men who feed the systems with data. The result: "The statements that come out often don't reflect reality. This is how stereotypes are reproduced." Her demand: more diversity is needed here.