DThe regulation of artificial intelligence (AI) is extremely important for the European economy. This is emphasized by Philipp Räther, who, as Group Chief Privacy Officer, is responsible for data protection at the Allianz insurance group. “We live in the age of digitalization, and this applies especially to banks and insurers whose business is particularly data-dependent,” he says in an interview with the FAZ. AI speeds up the processes significantly and also makes them cheaper. Räther also expects better product and service offerings in many areas through AI.
In his opinion, the spread of AI will have a significant impact on the economy. Some compare the impact to the computer revolution. However, Allianz's chief data protection officer still sees some pitfalls and snags in artificial intelligence (AI), which is why regulation is necessary. “In my opinion, the AI Act, i.e. the EU regulation, addresses this sufficiently,” praises Räther.
For him, the pressure to set a framework has increased due to generative AI, i.e. AI that learns itself and produces images and texts. Here, large data sets are compared and a result is derived based on probability. “This poses the risk of so-called hallucinations, i.e. false results,” admits Räther.
But the AI Act addresses these problems. The developers of the AI systems would have to be transparent about how they trained their systems. Users like Allianz now have the legal opportunity to ask whether copyrights have been taken into account in the AI system. According to Räther, the AI developer has so far been able to refuse this information with reference to trade secrets. With the AI Act he can no longer do that.
“Customer not just at the mercy of the machine”
EU regulation makes it possible to monitor AI instruments through systems that are also controlled via AI. “But people always remain the final authority who decides,” emphasizes Räther. According to him, if a customer complains because he feels he has been treated wrongly, a clerk must always be called in. “The customer will not just be at the mercy of the machine.”
The alliance could hardly understand politicians' concerns about the AI Act. Above all, the accusation that European providers were being discriminated against was aimed in the wrong direction for Räther. Rather, what matters in AI is trust in the systems: the AI Act could achieve this by better controlling the associated risks and leading to more responsible use.
Global regulatory framework
“The AI Act will help the AI industry and digitalization in all sectors.” He expects that the EU regulation will also set the future framework for AI regulation globally. This was also the case with the EU data protection regulation, which led to similar framework conditions in the USA and China.
In Räther's opinion, the EU regulation is aimed at the misbehavior of AI, i.e. at the wrong results. In addition, the AI Act creates more transparency in the systems used and prevents discrimination against groups of people so that they are not disadvantaged based on certain characteristics that are irrelevant to the insurance risk. It would be discriminatory if the name suggested a migrant background and the risk assessment was therefore worse.
Räther praises the fact that the AI Act does not provide too detailed requirements, but rather sets a sensible framework to address the most important risks and to control AI via people and for the benefit of people.
#Allianz #data #protection #officer #people #decide
