"This is the first International AI Safety Report. Following an interim publication in May 2024, a diverse group of 96 Artificial Intelligence (AI) experts contributed to this first full report, including an international Expert Advisory Panel nominated by 30 countries, the Organisation for Economic Co-operation and Development (OECD), the European Union (EU), and the United Nations (UN). The report aims to provide scientific information that will support informed policymaking. It does not recommend specific policies.
● The report is the work of independent experts. Led by the Chair, the independent experts writing this report collectively had full discretion over its content.
● While this report is concerned with AI risks and AI safety, AI also offers many potential benefits for people, businesses, and society. There are many types of AI, each with different benefits and risks. Most of the time, in most applications, AI helps individuals and organisations be more effective. But people around the world will only be able to fully enjoy AI’s many potential benefits safely if its risks are appropriately managed. This report focuses on identifying these risks and evaluating methods for mitigating them. It does not aim to comprehensively assess all possible societal impacts of AI, including its many potential benefits.
● The focus of the report is general-purpose AI. The report restricts its focus to a type of AI that has advanced particularly rapidly in recent years, and whose associated risks have been less studied and understood: general-purpose AI, or AI that can perform a wide variety of tasks. The analysis in this report focuses on the most advanced general-purpose AI systems at the time of writing, as well as future systems that might be even more capable.
● The report summarises the scientific evidence on three core questions: What can general-purpose AI do? What are risks associated with general-purpose AI? And what mitigation techniques are there against these risks?
● The stakes are high. We, the experts contributing to this report, continue to disagree on several questions, minor and major, around general-purpose AI capabilities, risks, and risk mitigations. But we consider this report essential for improving our collective understanding of this technology and its potential risks. We hope that the report will help the international community to move towards greater consensus about general-purpose AI and mitigate its risks more effectively, so that people can safely experience its many potential benefits. The stakes are high. We look forward to continuing this effort."
Great work from Yoshua Bengio and a large team of experts.
Also, I am very proud and motivated to see the MIT AI Risk Repository being cited in such an important paper.