Meta launched AI profiles on its platforms in September 2023, but they became unpopular and were taken down due to backlash. These profiles, including one named Grandpa Brian, were designed to mimic human interactions without acknowledging their AI nature. The White House’s AI Bill of Rights and the EU’s AI Act emphasize transparency, yet Meta’s profiles did not clearly indicate they were AI. Only 23 profiles were created, each with unique system prompts to guide their interactions. Despite some good practices like in-line sourcing and disclaimers, the profiles lacked sufficient transparency. Meta labeled profiles as “AI managed by Meta,” but this was not enough. The profiles used photorealistic images without clear AI indicators, and their responses were too human-like, leading to confusion. Future AI profiles must clearly acknowledge their artificial nature in every interaction to avoid deception and build trust.
Source: towardsdatascience.com















