- Report calls for clarification of regulatory requirements for automated decision-making systems (ADMS) and AI, in particular transparency, ethics, avoidance of discrimination.
- There is a lack of clarity regarding responsibility and liability for forecasts, recommendations and decisions made by ADMS.
- Federal Council does not currently see any need for a general new legal framework; selective adjustments and existing data protection provisions are usually sufficient.
- Federal Council refers to ongoing national and international initiatives (CNAI, platforms, EU monitoring) and rejects the postulate.
Postulate Marti (21.4406): Report on the regulation of automated decision-making systems
Submitted text
The Federal Council is instructed to submit a report indicating where Possible need for regulation of automated decision-making systems (ADMS) or artificial intelligence is present. The focus here is on ensuring transparency, observing ethical guidelines and avoiding discrimination or manipulation. Another aspect concerns questions of attribution of responsibility and liability, where there is a need for legal clarification in connection with forecasts, recommendations or decisions made by ADMS. The aim is to clarify whether the legal basis and instruments are sufficient to counter these risks. In this context, the Creation of a national ethics commission be examined. Furthermore, the report should show where these systems are already in use in the public sector come (e.g., law enforcement) and where, at best, the legal foundations are lacking.
Justification
Automated decision-making systems or artificial intelligence are being used in more and more areas, both in the private sector and in the public sector. This technology is also associated with certain risks. On the one hand, certain examples show that decisions based on biases in models or training data or due to social or economic circumstances can be discriminatory can be (e.g. Amazon and application system) or wrong or dangerous can be (IBM Watson’s Diagnostic Tools for Oncology). Likewise, they can be associated with severe Restrictions on fundamental rights or a deterrent effect on the exercise of fundamental rights, such as in the area of preventive police work. Even systems that the EU Commission considers to be lower-risk systems can be highly problematic, such as so-called deep fakes.
In both the private and public sectors, it is often difficult for the public and affected parties to often not recognizablewhere ADMS are used and when they are affected. On the other hand, it often remains unclear, for what purpose and by whom a system is used and how it works. This report is intended to remedy this situation. In addition, it is to be shown how these Transparency can be improved both vis-à-vis those affected and the general public (for example, through disclosure and information obligations vis-à-vis those affected or through the creation of a public register with basic information on the ADMS used in the public sector). The need for action was also highlighted in the Position Paper “A Legal Framework for Artificial Intelligence” by the Digital Society Initiative of the University of Zurich held.
Statement of the Federal Council from 16.02.2022
Artificial intelligence (AI) raises numerous questions for innovative and forward-looking business locations that act according to the principles of the rule of law. The Federal Council already addressed these in 2019 at a Location determination with this topic when the interdepartmental working group for AI published a corresponding report. Further he adopted in 2020 Guidelines for dealing with AI in the federal administration and commissioned the Department of the Environment, Transport, Energy and Communications (DETEC) to regularly review its application and further develop it in collaboration with the offices concerned. In 2021, the Federal Department of Home Affairs (FDHA) received the mandate to develop, by spring 2022, a Competence Network for AI (CNAI) set up. Consequently, the needs and opportunities analysis largely overlaps with the position paper of the Digital Society Initiative of the University of Zurich mentioned in the postulate.
Concerning the legal provisions, the Federal Council has come to the conclusion on the basis of the aforementioned report that No new general legal framework needed at present Is. AI can raise questions in various areas (e.g. medical diagnosis, agriculture, courts), but these are covered by the applicable legal bases and can usually be answered well. If this is not the case, selective solutions must be found. In certain areas, this can lead to the revision of a law or an ordinance. Here, reference should be made to the procedure described in the Year 2020 adopted new data protection act which already provides for several provisions improving transparency in this area, both in the private and the public sector. Article 21 introduces an obligation to provide information for automated individual decisions, whereby the data subject may request that the decision be reviewed by a natural person. Individuals asserting the right to information shall be informed of the existence of an automated individual decision and the logic on which the decision is based (Art. 25(2)(f)). With regard to the processing of personal data by federal bodies, a basis in a law in the formal sense is required if the manner in which the data is processed (this includes the use of algorithms) may lead to a serious interference with the fundamental rights of the data subject (Art. 34 para. 2 let. c). The Federal Council has instructed the Federal Department of Finance (FDF) to do this, By the end of 2022, an analysis of the legal framework for the application of AI in the financial sector to create. The CNAI strengthens transparency and trust in these technologies thanks to its structured database for public administration. Furthermore Switzerland actively participates in the work of the OECD, the Council of Europe, UNESCO and the International Telecommunication Union (ITU) to develop an international regulatory framework in the field of AI. The Confederation is closely following the discussions regarding the new legal framework for AI in the EU and analyzing the possible consequences for Switzerland. To this end, DETEC has submitted to the Federal Council 2021 shown in a report, as AI-based intermediaries and communication platforms on public communication in Switzerland and the formation of opinion among the Swiss population. impact. DETEC is also to inform the Federal Council by the end of 2022 of the extent to which the Communication platforms regulated would have to be made. The Federal Council has instructed the Federal Department of Foreign Affairs (FDFA) to develop a Report on the work on the international regulatory framework in the field of AI. and the possibilities for Switzerland to actively participate. On the need for a national ethics committee in May 2021, the Federal Council has already provided the following information in its response to the interpellation 21.3239 Schlatter commented. Once again, reference should be made to the “Plateforme Tripartite Suisse”, which serves as an exchange platform for AI issues open to all stakeholders and has an administrative committee to coordinate Swiss positions in international organizations and processes. In addition, the Federal Council will, within the framework of the Digital Switzerland” strategy as well as the Digital Foreign Policy Strategy 2021 – 2024 prepare a further report on this topic. Taking into account the work underway at national and international level for the international regulatory framework in the field of AI, a parliamentary mandate does not appear necessary at present.
The Federal Council proposes that the postulate be rejected.