UNESCO publishes policy paper on AI foundation models

UNESCO is weighing in on AI ethics with its own recommendations and tools, including the Readiness Assessment Methodology. It's an ex-ante assessment for governments, companies, and other organizations to design, develop, deploy, and procure AI systems ethically and in accordance with human rights and fundamental freedoms.

OpenAI could take several steps to mitigate some of the identified risks, such as fully disclosing the dataset used to train GPT-4 and ensuring that ChatGPT provides references to support any factual claims it makes in its responses.

 

From the paper