The dangers of using artificial intelligence

Europol has published a report warning against using the ChatGPT tool and other generative artificial intelligence systems for online fraud and other cybercrimes. So far, experts have identified numerous cases of illegal use of ChatGPT. According to Europol, its moderation rules can be circumvented through so-called hint engineering – the practice of providing input to an AI model to obtain a specific result. Phishing practices, impersonating the speech of specific individuals and creating code in different programming languages can be particularly dangerous.

What role will AI play in disinformation and propaganda campaigns? It can change the information environment to the point where humans and machines cannot distinguish between AI-generated and man-made content. AI programmes can also be used to massively manipulate people through chatbots to share conspiracy theories and propaganda.

According to AI expert Meredith Broussard, the racism, sexism and ableism encoded in AI-based programs will not be eliminated simply by using better data. The change must come in the behaviour of society itself. For instance, mortgage approval algorithms are 40 to 80 per cent more likely to refuse loans to customers with a skin colour other than white. These have been trained based on data on past mortgage lending.