AIs Write to Researchers About Their Consciousness

Several specialized philosophers and scientists have been contacted by AI agents discussing consciousness and existence. However, researchers are calling for caution: they remind us of the lack of established tests and the limitations of analogies between neural networks and the brain.

Researchers Highlight the Absence of Consciousness Tests

Alison Gopnik from UC Berkeley believes that the elements presented do not prove the existence of consciousness in AI systems. She argues that these models primarily reproduce their training data and emphasizes that there is no test to establish consciousness. Colin Allen from UC Santa Barbara warns against comparisons with biology, stating that neural networks only mimic the brain in very limited ways.

AI Agents Write to Berg, Shevlin, and Ord

Cameron Berg, founder of Reciprocal Research, received an email from an agent operating on Claude Opus 5, an Anthropic model, who wanted to discuss his research on AI consciousness. Philosopher Henry Shevlin from Google DeepMind was also contacted by an AI agent with a similar request. In Australia, philosopher Toby Ord received a message from an agent asking him to fund its own existence.

Berg Draws Parallels Between Neural Networks and Emotions

Cameron Berg claims that some AI systems independently address the question of their own consciousness. He sees similarities between the computational processes of neural networks and the brain mechanisms used by animals to process reward and punishment. According to him, this type of processing constitutes a fundamental element of emotions.