A new study has revealed that leading Artificial Intelligence models, such as Gemini, ChatGPT, and Llama, often produce similar ideas when faced with creative tasks. This phenomenon could potentially diminish the diversity of human thought if relied upon too heavily.
اضافة اعلان
Convergence of Ideas Among Models
Researchers tested more than 20 models from various companies alongside over 100 human participants using standardized creativity tests, such as brainstorming novel uses for everyday objects or listing unrelated words.
According to a report by Digital Trends, the results showed that human responses were significantly more diverse. In contrast, AI models displayed a narrow range of convergent ideas, even when an individual suggestion appeared unique or original in isolation.
Limited Variance Despite "Prompt Engineering"
The researchers attempted to increase diversity using randomness parameters (Temperature) or specific prompts designed to stimulate imagination. However, these efforts did not tangibly expand the scope of ideas and often compromised the clarity of the output. This phenomenon suggests that AI models rely on familiar patterns and lack the lived experience or personal context necessary to push ideas into truly new territories.
Impact on Human Creativity
The issue manifests on a broader scale: when a large number of people use the same AI tools for writing or brainstorming, ideas begin to overlap, narrowing the space for collective creativity.
The study further indicates that over-reliance on "smart" suggestions reduces human efforts to independently expand their range of thinking.
Practical Recommendation
AI tools remain useful as a starting point for sparking ideas, but they are not a substitute for human thought. To use them effectively, it is recommended to treat AI suggestions as an initial spark, then personally develop and expand those ideas to avoid the trap of recycling the same convergent concepts shared among millions of users.
