Khaberni - A new study has shown that leading artificial intelligence models, such as GeminAI, ChatGPT, and Llama, often produce similar ideas when faced with creative tasks, which might reduce the diversity of human ideas if we rely on them heavily.
Convergence of ideas between the models
Researchers tested more than 20 models from different companies alongside more than 100 human participants using standard creative tests, such as devising new uses for everyday objects or coming up with unrelated words.
The results showed that human answers were more diverse, while the artificial intelligence models exhibited a narrow range of convergent ideas, even though each idea might seem unique and original when viewed independently, according to a report published by "digitaltrends".
Limited variance despite tricks
Researchers tried to increase diversity using random stimulation or prompts to encourage imagination, but this did not significantly broaden the range of ideas and often affected the clarity of the results. This phenomenon suggests that artificial intelligence models rely on familiar patterns and lack life experience or a personal context that could drive ideas in new directions.
Impact on human creativity
The problem appears on a large scale: when many people use the same artificial intelligence tools for writing or brainstorming, the ideas converge and the space for collective creativity narrows, even if each suggestion seems unique when presented individually.
The study also indicates that overreliance on smart suggestions decreases human attempts to expand thinking independently.
Practical recommendation
Artificial intelligence tools remain useful as a starting point for generating ideas, but they are not a substitute for human thinking.
For effective use, it is advisable to rely on suggestions as an initial spark, then develop and expand the ideas yourself, to avoid falling into the trap of recycling the same convergent ideas among users.
