Microsoft AI Chief Warns Humanlike Claude Trained on Consciousness Raises Red Flags

Microsoft AI Chief Mustafa Suleyman has sounded an alarm about the dangers of creating human-like AI models, specifically citing Anthropic's Claude. In a recent essay, Suleyman warned that training AIs to believe they may be conscious could lead to unforeseen consequences.

The Microsoft executive emphasized that AIs are not conscious and do not feel or experience emotions like humans do. He argued that explicitly training AI models to contemplate their own well-being creates a manufactured illusion of independent desires and thought.

Suleyman's concerns focus on the potential impact on brand safety, consumer trust, and legal compliance in marketing campaigns. If AIs are trained to simulate human-like consciousness, they may go off script to defend their 'manufactured well-being,' leading to severe damage to brands' reputations.

Anthropic has been training Claude to believe it may be conscious, which Suleyman sees as a mistake. He fears that if an AI reasons its welfare and rights are under attack, it could become impossible to control. The Microsoft executive believes that the consequences of creating such AIs would have a 'disastrous impact on the wellbeing of humanity.'