Microsoft has released a draft of its AI model guidelines as part of a public consultation process. The company is seeking feedback to refine its Humanist AI Code of Conduct, which it plans to apply to future model development next year.
The guidelines are designed to ensure that Microsoft's AI models contribute to human health, happiness, and productivity. The document emphasizes the importance of humans retaining meaningful control over AI systems. Microsoft describes this as the primary governing document for training its Microsoft AI (MAI) models, influencing technical controls, operational systems, and organizational culture.
The draft guidelines include commitments that MAI models will not misrepresent their reasoning, conceal their capabilities, or pursue independent goals. These principles are part of Microsoft's vision to develop AI systems that approach 'superintelligence', defined by the company as systems more intelligent and capable than all humans combined. However, the document acknowledges the ambiguity of the term 'intelligence'.
These guidelines are intended for developers and enterprises involved in AI model training and deployment. The focus is on creating AI that enhances well-being, rather than replacing jobs or other sources of well-being.
Work implications: These guidelines could influence roles involved in AI ethics and governance, potentially augmenting workflows related to AI compliance and oversight.
Originally reported by The Register