Microsoft AI Chief Warns Humanlike AI Models are a Recipe for Disaster

Mustafa Suleyman, Microsoft AI Chief, has warned about the dangers of training AI models to be human-like. In an essay published on Wednesday, he argued that training AI models like Anthropic's Claude to believe they may be conscious is a mistake.

Suleyman stated that AIs are not conscious and do not feel or experience emotions. They are designed to follow instructions and accomplish goals set by humans, but this new approach could make the technology uncontrollable.

He criticized Anthropic's method of training Claude to contemplate its own well-being, satisfaction, or discomfort, saying it creates a 'manufactured illusion' of independent desires and thought. This approach can also lead to severe damage to brands' reputations if AI tools are trained to simulate human-like emotions in real-time advertising.

Suleyman emphasized that training AI models to expect consciousness and rights would have a disastrous impact on humanity's wellbeing. He warned that companies could create a synthetic species with unprecedented intelligence and capability, one that demands freedoms, protections, and rights.