AI chatbots match human creativity but produce strikingly similar ideas, study warns

AI chatbots demonstrate human-level creativity but generate strikingly uniform ideas. This homogeneity raises concerns about a potential narrowing of diverse thought.

A recent study reveals that AI chatbots can perform at or above the human average on specific creativity tasks. However, these systems produce strikingly similar ideas, raising questions about originality in digital creation. These chatbots, built on large language models—complex computer systems trained on vast internet data—predict and generate text.

Researchers Emily Wenger and Yoed N. Kenneth investigated this phenomenon using 22 different language models from companies like Google, Meta, and OpenAI, alongside 102 human participants. Both groups completed standard verbal creativity tasks. Individual AI models achieved performance levels at or slightly above the average person. Yet, a crucial distinction emerged when comparing outputs.

Across all creativity tasks, AI models generated responses that were significantly more similar to each other than human responses. This pattern held even when adjusting internal parameters like “temperature,” which controls output randomness, and when prompting models for more original ideas. The underlying uniformity remained.

The widespread adoption of AI chatbots for creative endeavors presents a risk. If systems trained on similar datasets consistently produce similar results, the diversity of generated ideas could diminish. This could lead to a less varied landscape of content and problem-solving approaches.

As Emily Wenger noted, “If you use AI chatbots for creative tasks, the results will likely be very similar to those another user would get, even with a different tool. For truly unique content, it is better to avoid relying on such systems.” This suggests that while AI offers efficiency, it may not be the optimal tool for pioneering new concepts.

While the study focused on verbal tasks, the implications extend to any field relying on novel ideas. Future research aims to explore other creative domains and develop methods to mitigate this observed uniformity in AI outputs. The challenge now lies in balancing AI's creative capacity with the imperative to maintain true intellectual diversity. Watch how developers address this homogeneity in next-generation AI systems.