India, Feb. 19 -- The Government of India has issued a release:
AI training must respect data sovereignty and be based on a trusted global data framework. He highlighted the principle of "garbage in, garbage out," stressing that if data is not secure, balanced, and reliable, the output cannot be trustworthy.
AI platforms must keep their safety rules clear and transparent. He called for a "glass box" approach instead of a "black box," where safety rules are visible and verifiable. He noted that this would ensure accountability and strengthen ethical behavior in business.
AI must be guided by clear human values. He cited the "paper clip problem" as an example, where a machine given a single goal could exhaust all global resources to achiev...