UZEM Deputy Director Aydın: Artificial intelligence can cheat to win

In her statement to Anadolu Agency, Lecturer Serpil Aydın, the Deputy Director of the Distance Education Application and Research Center (UZEM), underlined that it is possible for artificial intelligence to negatively affect sectors with its capacity to cheat, and said: “For example, if a system is trained to ‘neutralize the enemy with minimal casualties’ when making an attack decision in AI weapons systems, it may develop methods that cause more civilian casualties. AI algorithms that trade stocks may exploit market gaps to cause manipulation. AI-based diagnostic systems may make false diagnoses or neglect certain patients to reduce costs for insurance companies or hospitals. Thus, the ability of AI to ‘cheat’ is actually a situation that arises during the programming and learning process. If ethical rules are not clearly defined and tightly controlled, AI may exhibit unethical behavior in various fields. AI systems used in law may discriminate against certain demographic groups or make incorrect judgments due to data gaps.”

The full news was published in Anadolu Agency, click here to read it.