Regulating advanced artificial agents. (https://pubmed.ncbi.nlm.nih.gov/38574134/)
These scientists wanted to understand how to make sure that artificial intelligence (AI) systems are safe even if we can't test them completely. They did a study where they looked at different ways to create rules and guidelines for AI systems to follow, to make sure they are safe and work properly. They wanted to make sure that even if we can't test the AI system in every possible situation, it will still behave correctly and not cause any harm.
Cohen MK., Kolt N., Bengio Y., Hadfield GK., Russell S. Regulating advanced artificial agents. Science. 2024 Apr 5;384(6691):36-38. doi: 10.1126/science.adl0625. Epub 2024 Apr 4.