CreativityNeuro: Steering Language Model Weights to Improve Divergent Thinking and Reduce Mode Collapse

Researchers have introduced a method called CreativityNeuro designed to increase divergent thinking in large language models by adjusting internal weight parameters. This approach aims to counter the tendency of models to produce repetitive, overly similar outputs known as the artificial hivemind effect. By modifying how these models generate open-ended responses, the technique seeks to improve the variety and creative potential of AI-generated content.