13.05.2024
| Research |
PhD-Student at DSC Sana Hassan Imam Presents Paper on Generative AI Model
Sana presented her paper on a generative AI model at the ACM CHI Generative AI Workshop 2024 on the 12th of May. By this she brought international recognition to the research at DSC.
Sana Hassan Imam is a PhD-Student at DSC as part of a collaborative project with the faculty of Economics. Her paper with the title “How Can Generative AI Curate the User Creativity on an Idea Crowdsourcing Platform?” discusses how generative AI can enhance user creativity on a so-called
idea crowdsourcing platform by evaluating the uniqueness, diversity, and feasibility of ideas, and providing targeted prompt designs to improve the quality of those ideas with the help of a proposed model.
Moreover, the paper showcases the limitations of traditional methods: Traditional idea-generation methods often limit novelty and diversity, requiring significant human creativity which can be hard to measure. The generative AI model mainly works in two steps:
Step 1: Evaluates creativity scores of user ideas based on uniqueness, diversity, and feasibility, using feedback from peers and managers
Step 2: Provides tailored prompts to users to stimulate, enhance, and refine their ideas based on the calculated creativity scores
To implement this model, scores for creative engagement, idea uniqueness, diversity, and flexibility are calculated. Using these scores, a novel prompt-designing mechanism has been developed. Instead of providing direct suggestions, the model uses prompts to encourage creativity without creating dependency on artificial intelligence. Examples of how to increase novelty include adding random words, using “What if”- scenarios to stimulate divergent thinking, and encouraging peer and manager feedback to improve feasibility. This feasibility is assessed through feedback summaries and different prompt designs are suggested to enhance user creativity based on these scores.
To bring this model even further, a synthetic data example, more specifically a user-generated idea for a “Smart Umbrella project”, is brought up to illustrate the model’s idea. The example aims to incorporate a dynamic light show using LED lights along the canopy’s rim. Its prompts were designed to make the idea more practical by focusing on user experience, core functionality, cost, and accessibility.
In summary, this model aids users in refining and enhancing their ideas by stimulating creative thinking. The iterative process of idea enhancement improves the quality and creativity of user-generated content. Additionally, this approach helps platform managers identify high-potential ideas with high uniqueness, diversity, and feasibility scores that may not have received enough votes.
The paper concludes by stressing the “holistic framework for enhancing user creative engagement in idea crowdsourcing using generative AI as a thought provoker and co-creator”.
Updated by: Jennifer Nüchter
« back
13.05.2024 | Research
PhD-Student at DSC Sana Hassan Imam Presents Paper on Generative AI Model
Sana presented her paper on a generative AI model at the ACM CHI Generative AI Workshop 2024 on the 12th of May. By this she brought international recognition to the research at DSC.
Sana Hassan Imam is a PhD-Student at DSC as part of a collaborative project with the faculty of Economics. Her paper with the title “How Can Generative AI Curate the User Creativity on an Idea Crowdsourcing Platform?” discusses how generative AI can enhance user creativity on a so-called
idea crowdsourcing platform by evaluating the uniqueness, diversity, and feasibility of ideas, and providing targeted prompt designs to improve the quality of those ideas with the help of a proposed model.
Moreover, the paper showcases the limitations of traditional methods: Traditional idea-generation methods often limit novelty and diversity, requiring significant human creativity which can be hard to measure. The generative AI model mainly works in two steps:
Step 1: Evaluates creativity scores of user ideas based on uniqueness, diversity, and feasibility, using feedback from peers and managers
Step 2: Provides tailored prompts to users to stimulate, enhance, and refine their ideas based on the calculated creativity scores
To implement this model, scores for creative engagement, idea uniqueness, diversity, and flexibility are calculated. Using these scores, a novel prompt-designing mechanism has been developed. Instead of providing direct suggestions, the model uses prompts to encourage creativity without creating dependency on artificial intelligence. Examples of how to increase novelty include adding random words, using “What if”- scenarios to stimulate divergent thinking, and encouraging peer and manager feedback to improve feasibility. This feasibility is assessed through feedback summaries and different prompt designs are suggested to enhance user creativity based on these scores.
To bring this model even further, a synthetic data example, more specifically a user-generated idea for a “Smart Umbrella project”, is brought up to illustrate the model’s idea. The example aims to incorporate a dynamic light show using LED lights along the canopy’s rim. Its prompts were designed to make the idea more practical by focusing on user experience, core functionality, cost, and accessibility.
In summary, this model aids users in refining and enhancing their ideas by stimulating creative thinking. The iterative process of idea enhancement improves the quality and creativity of user-generated content. Additionally, this approach helps platform managers identify high-potential ideas with high uniqueness, diversity, and feasibility scores that may not have received enough votes.
The paper concludes by stressing the “holistic framework for enhancing user creative engagement in idea crowdsourcing using generative AI as a thought provoker and co-creator”.
Author: Jennifer Nüchter
For further questions, please contact:
Sana Hassan Imam
Doctoral Candidate
+49 (421) 218 - 63954
sanahassan@uni-bremen.de
« back