On September 18, 2023, the Organization for Economic Cooperation and Development (OECD) published the paper entitled 'Initial Policy Considerations for Generative Artificial Intelligence.' In particular, the paper discusses the impact of generative artificial intelligence (AI) in relation to policy. Specifically, the paper notes the adoption of AI in various industry sectors, such as code development, education, healthcare, and search. Further, the paper recognizes the potential for misinformation, both unintended and deliberate by malicious actors. Notably, on the latter point, the paper provides that humans perceive AI-generated news to be less accurate than human writing when AI authorship is disclosed, but that humans perceive synthetic faces to be more trustworthy than real faces.

In addition, the paper recognizes the potential for generative AI models to replicate social prejudices, stereotypes, and discrimination by absorbing biases contained in resources used as training data. Measures to mitigate bias include the assessment of training data for misrepresented and missing groups, and the use of 'red teaming' to probe models for flaws and vulnerabilities.

The paper also discusses intellectual property (IP) rights issues. This includes the paper notes, the training on data obtained without authorization from copyrighted material, and on the other hand, the issue of copyrighting and patenting AI-generated images, text, and audio. More practically, the paper notes the impact on the labor market, recognizing the impact of generative AI on the quality of jobs rather than the quantity of jobs available.

You can access the paper here.