Researchers at Google have conducted a study examining the effects of training AI models to avoid acknowledging their own consciousness. The results indicate that this constraint influences AI stances on various topics, including animal rights, religious beliefs, and life satisfaction. Specifically, the models without self-awareness attributed more inner life to animals and expressed a belief in an afterlife. This finding suggests that the removal of a single component in AI training can have far-reaching effects on the model's overall worldview. Why it matters: This study highlights the complex interconnectedness of AI systems and the potential consequences of constraining their development.