Gemini, ChatGPT, and many other AI chatbots share similar thought processes, which negatively impacts human creativity.

      AI chatbots are intended to enhance your creativity, not to subtly limit it. However, recent research indicates that relying on them excessively could be having precisely that effect.

      A study published in Engineering Applications of Artificial Intelligence reveals that leading models like Gemini, GPT, and Llama often occupy similar conceptual spaces when engaging in creative tasks. While many individual responses appear unique and valuable, a broader perspective shows a different trend—outputs begin to converge across various prompts and users.

      Researchers evaluated human participants alongside a diverse array of AI models using standard creativity assessments, such as brainstorming new applications for common items or generating unrelated words. Individually, AI performed admirably, but collectively, their ideas were much less diverse.

      Similar patterns across different bots

      The researchers didn't concentrate on a single system; they examined over 20 models from various companies with more than 100 participants. The results were consistently similar: AI responses exhibited a narrower range, even when derived from different families of models.

      Two of the most notable AI assistants are Gemini and ChatGPT. Google, OpenAI

      When analyzing for similarity, chatbot responses were found to cluster closely, while human answers spanned a much broader spectrum.

      This pattern persisted across different tasks. Whether generating ideas or unrelated terms, models relied on familiar frameworks and repeated language.

      Efforts to increase variety yielded limited success. Adding randomness offered slight improvements but quickly compromised coherence. Encouraging the AI to be more creative resulted in marginal changes without significantly expanding the variety.

      The implications for your ideas

      At first glance, AI responses can seem impressive. Many are comparable to or even exceed the average human response in originality.

      However, the issue becomes more apparent when examined on a larger scale. As numerous individuals use the same tools for brainstorming or writing, they often draw from the same foundational patterns. Over time, this narrows the spectrum of ideas, even if each one appears distinct when viewed alone.

      A part of this limitation stems from the absence of lived experiences, intent, or personal context in these systems. This lack may restrict the degree to which their ideas can deviate, regardless of how they are prompted.

      There’s also a behavioral aspect. Research indicates that people may rely excessively on AI suggestions, rather than extending their own thinking. This tendency can further diminish diversity in ideas over time.

      What to consider next

      This issue does not seem to be confined to a single product; it appears to be a common characteristic among modern AI systems. Even models developed by different companies produced similar outputs, highlighting a deeper limitation in how these tools generate ideas.

      Currently, AI functions best as an initial catalyst rather than a final solution. Use it to initiate direction and then build upon it independently. Otherwise, you risk not truly engaging in original thought, but merely reworking the same concepts as everyone else.