Gemini, ChatGPT and most other AI chatbots think alike, and it’s bad for human creativity
A new study finds that leading AI chatbots — including Gemini, GPT, and Llama — produce strikingly similar answers when tackling creative tasks. While individual responses may seem original, researchers discovered that across many prompts and users, the models converge on the same conceptual space. Testing more than 20 AI systems against over 100 humans, the study shows that human ideas are far more diverse, while AI outputs cluster tightly together. Attempts to force more variety through randomness or creative prompting had limited effect. The concern is that heavy reliance on AI could narrow human creativity over time, as people draw from the same underlying patterns.