Microsoft Publishes Draft Document Outlining Rules for AI Training

Microsoft published a document that outlines the intended behavior and values of its AI models, summarizing the company's approach to training and operating them. Microsoft noted that the document is in the works and currently, it's not implemented. The company shared it for public consideration and feedback and will publish the revised version at the end of the year.

Microsoft it working towards the concept of Humanist Superintelligence (HSI), set out last year. Humanist Superintelligence is an "incredibly advanced AI capabilities that always work for people and in the service of humanity more generally." The overall idea is that the technology should serve humanity and not exceed human control. If a technology can't remain within human control, it should be rejected. The document also states that artificial intelligence is artificial – it should not have legal personhood or designed to be perceived as a person. The goal of the technology is to "drive better living standards and human flourishing" and develop human potential and agency.

According to the document, AI should embrace the diversity of traditions, contexts, and forms of flourishing that humanity has, while not being neutral toward harm – there's an opinion that this point is controversial, as values can conflict, and there's an open question what should be done in that case.

The code also states that the technology should not violate individual human dignity, development, autonomy, or security, and recognize when somene tries to do it. The document also says that having access to information doesn't automatically mean the model has permission to reproduce it.

Human control also expresses in the way how AI models communicate and says that they shouln't communicate in neuralese or any form beyond simple human understanding.

Microsoft has worked on the code, attracting experts in AI, law, ethics, philosophy, linguistics, and public policy, business leaders from the industry, and focus groups of members from public.

The risks of increasingly autonomous AI systems have also become more real. For example, Hugging Face recentlty discovered an intrusion into its production infrastructure driven by an autonomous AI agent system.

Learn the details about Microsoft's Code here and read its text by clicking this link. 

Also, subscribe to our Newsletter, join our 80 Level Talent platform, Discord, and follow us on Twitter, LinkedIn, Telegram, and Instagram, where we share breakdowns, the latest news, awesome artworks, and more.

Are you a fan of what we do here at 80 Level? Then make sure to set us as a Preferred Source on Google to see more of our content in your feed.