AI Is Learning to Manipulate Us, and We Don't Know Exactly How


Dezfouli points out that whether these technologies are used for good or ill depends on how responsibly we design and deploy them. In an attempt to ensure good outcomes, CSIRO and the Australian government developed an ethics framework for AI in government and industry. These (voluntary) principles include much of what you might expect, like "AI systems should respect and uphold privacy rights and data protection." Another tenant says that transparency and responsible disclosure are crucial, so that people can understand when their choices are being guided and find out when an AI system is engaging with them.