The Consultation Summary for UNESCO’s Recommendations on the Ethics of Artificial Intelligence

By August 31, 2020Uncategorized

AI Global was recently asked to run a consultation as part of UNESCO’s series of public consultations for their Draft Recommendations on Ethics of Artificial Intelligence

 

UNESCO has done a tremendous job connecting all of the key reports on AI ethics which have emerged over the past several years and through an international group of AI experts have consolidated these concepts into a comprehensive set of recommendations for governments, academic institutions, and companies to adopt. 

 

As we’ve seen in past reports about responsible and ethical AI, the governance of these systems is imperative.  Given our focus on AI regulation, we focused on reviewing these recommendations during our consultation. We were specifically interested in the downstream impacts that new AI regulations would have on two key groups: small to medium business and education professionals. 

 

The key takeaways were: 

 

  • A single framework for responsible and ethical AI is needed similar to UNESCO’s Universal Declaration on Bioethics and Human Rights which has helped to shape scientific testing as well as pharmaceutical regulation. This document should aim to provide a similar framework for the use of AI.  This will help businesses prepare for future regulations and will help academic institutions know how to shape current and future programming. 
  • There is a strong strong consensus for increased AI Governance. All recommendations relating to governance, regulation, and certification should be clear, direct, and use the strongest language possible.  
  • Prior to implementation, the purpose and scope of these regulations should be consulted with the impacted communities, both companies and organizations responsible for implementing these new rules as well as the people who these rules are supposed to be protecting. 
  • Stronger emphasis is required on the role and use of data and how it is used in AI systems.
  • All recommendations should be clear on what the desired objective is, as the implementation will vary greatly, several of the recommendations could use more precision. 
  • There is a need to ensure these recommendations have enough specificity to avoid being used for ethics washing. 

 

If you’re interested, the full report can be found here