Could AI be less blunt?
Buijsman, author of the book Artificial intelligence, increasingly intelligentFirst and foremost, he explains exactly what AI is: “It is computer systems that try to use pattern recognition to try to do the tasks that people would normally do. So you show them millions of examples of what they need to solve and then you hope the computer learns what It must be done.”
Computing power allows computers to calculate much more than a human can. But a computer can also make mistakes. For example, the anti-cheating program did not recognize students of color. Puigsmann believes that this is probably mainly due to human error: “When training the algorithm, examples of mainly white men and white women were used, which is why computers did not recognize students of color. Men and herein lies the flaw.”
So the companies behind the self-learning program should be aware of the potential errors. Moreover, the program is not always watertight, because according to Boijsmann it does not exist. Ultimately, errors are not the fault of the computer, but the responsibility of the person behind the software, he says: “It is up to us to handle this properly. We must combat discrimination and deal with errors in a humane manner.” According to him, programmers and software developers must be on the Particularly aware of the possibility of their AI being discriminatory and trying to prevent it.
“Travel enthusiast. Alcohol lover. Friendly entrepreneur. Coffeeaholic. Award-winning writer.”