AI Now calls for governmental regulation of AI

In its third annual report on the state of artificial intelligence (AI), the USA-based AI Now Institute calls for governments to regulate AI ‘by expanding the powers of the sector-specific agencies to oversee, audit, and monitor AI technologies by domain’. The AI Now 2018 Report, which focuses on the US context, draws attention to a lack of meaningful accountability and oversight (including with regards to responsibility, liability and due process) when it comes to the development and use of AI technologies. To tackle this and other issues, the institute makes nine other recommendations in addition to the need for governmental regulation. For example, because internal governance structures at many technology companies are found to be failing in ensuring accountability for AI systems, new approaches to governance are recommended as a matter of urgency. Another recommendation says that AI companies should waive trade secrecy and other legal claims that contribute to making AI systems opaque and unaccountable. Moreover, consumer protection agencies should apply ‘truth-in-advertising’ laws to AI products and services, to avoid the risks that ‘marketing promises’ can cause both individuals and commercial customers. For meaningful fairness, accountability and transparency in AI, a better understanding and tracking of the full supply chain of AI systems are needed.