BackThe European Ombudsman has made a series of suggestions to help the European Commission ensure the use of artificial intelligence (AI) in its decision making remains accountable.
The European Ombudsman welcomed that the Commission has publicly set out its approach to using AI as well as taken measures to ensure its use of AI systems complies with the recently-adopted AI Act. As many of the Act's requirements apply to AI classified as 'high-risk', the European Ombudsman stressed that it was important to make sure all AI systems that support the Commission’s decision making produce accurate results, enable it to provide reasons for its decisions, and remain under effective human oversight. She asked the Commission to consider adopting a voluntary code of conduct for systems not deemed high risk to better ensure their use is in line with the principles of good administration.
The European Ombudsman further noted that, while AI systems can be used as tools to support decision-making processes, decision making should not be delegated to these systems. She proposed that civil servants responsible for human oversight of AI receive adequate training on the risks and errors produced by AI systems and on how to critically examine and verify AI-produced results. Internal working procedures should also make clear that staff responsible for the oversight of AI systems are encouraged to depart from the systems’ suggestions in specific situations.
The European Ombudsman also underlined the importance of transparency when it comes to ensuring public trust that the EU administration will use AI responsibility. She encouraged the Commission to reflect on what information it will make publicly available around future AI projects and to consider carrying out public consultations on the possible adoption or development of powerful new AI tools.
For AI systems that produce decisions or replies to the public, the European Ombudsman proposed the Commission examine the need to proactively inform the public about the fact that a reply was prepared with the assistance of AI as well as about the nature of the AI assistance provided.
The European Ombudsman's proposals are based on an information-gathering exercise, for which the Commission had provided her with details about its use of AI in three particular areas—the analysis of public feedback, the discovery of potential infringements of EU competition rules, and complaint handling.
Kindly click here to access the Ombudsman's suggestions.
Source: European Ombudsman
