Artificial intelligence is an important driver of digital change and is being used in more and more areas. "Citizens, consumers and employees should be informed as transparently as possible about the use of AI in their dealings with the state, companies or their employer," according to the study published on Wednesday.
For example, the use of AI systems should be made clear and easily recognizable so that those affected know that they are interacting with an AI system and not with a human being.
Human testing necessary
The authors also write that important decisions that affect people should not be left to an AI system without weighing up the pros and cons. When it comes to personal issues, the result of the system usually has to be checked and justified by a human.
In contrast to conventional computers, many AI systems are able to recognize hidden regularities and suggest solutions thanks to training on large data sets. Advanced AI systems can also perform tasks that are currently the responsibility of well-trained professionals. AI is used in medicine, for example, to assess ultrasound, magnetic resonance or X-ray images.
According to the authors, the opportunities and risks of AI cannot be assessed in general terms due to the broad field of application. A general "AI law" would therefore not be expedient. Instead, problems and undesirable developments caused by AI should be regulated in existing laws or ordinances or, if necessary, with voluntary measures.
The study examines the use of AI in shopping, in traditional and new media, in the world of work, in education and research and in public authorities.
Requirements for government agencies
The use of AI places particular demands on state actors. However, government agencies in Switzerland still rarely use AI at present. They must ensure that the fundamental rights of citizens are not violated.
According to the study, AI may only be used in administration and jurisdiction according to clear criteria. The administration must also ensure that the data on which the AI systems are based is correct and complete.
With regard to the world of work, the authors also see an opportunity in the use of AI: thanks to efficiency gains, AI could help to reduce personal working hours. In order to make this possible and avoid feared job losses or social upheaval, education and training in AI must be promoted.
Ethical standards
In education, AI can be used to analyze the skills of learners and suggest educational options. "This involves collecting and generating large amounts of data about the individuals concerned," write the authors. "Educational institutions must therefore ensure that students are not disadvantaged later on as a result."
Imparting knowledge about artificial intelligence is another aspect for the authors. Universities should clarify how an understanding of the capabilities and limitations of AI systems can be promoted. It should also be the task of universities to develop ethical standards for the research and use of AI.
Difficult balance
Anyone who orders goods online inevitably comes into contact with AI. According to the authors, personalized shopping has become a matter of course, the information technology processes work in the background and have hardly been questioned to date.
According to the experts surveyed as part of the study, increasing awareness among users of the value of their own data is a pressing concern. However, it is difficult for companies to reconcile personalization and data protection.
Open data and control options for customers, transparent disclosure of the use of AI and the right to data deletion were seen by the experts surveyed as promising measures to promote customer trust.
Federal government to raise awareness of false reports
Algorithms and artificial intelligence are also increasingly helping to determine which news and content users are shown on media platforms. According to the study, platform operators are called upon to make it clear which messages are shown due to the use of AI systems.
The federal government should raise awareness of false reports that are created with the help of AI. At the same time, the security authorities - under parliamentary control - should be able to recognize political campaigns operating with false reports and inform the population about them.
The study was carried out by researchers from the Digital Society Initiative of the University of Zurich, the Technology and Society department of Empa (Swiss Federal Laboratories for Materials Science and Technology) and the Institute for Technology Assessment of the Austrian Academy of Sciences. The TA-Swiss Foundation assesses the consequences of technology and provides the basis for comprehensively weighing up the opportunities and risks. (SDA)