AI-Augmented Brainwriting: Investigating the use of LLMs in group ideation
Project Overview
The document examines the incorporation of large language models (LLMs) in educational settings, particularly focusing on their application in the Brainwriting process within a tangible interaction design course. The study reveals that the integration of LLMs significantly enhances both idea generation and evaluation, resulting in more fruitful creative project outcomes. Despite these advantages, it also identifies challenges, including the potential for redundancy in the outputs generated by LLMs and the critical need for effective prompt engineering to optimize their use. Overall, the findings indicate a positive outlook for human-AI collaboration in education, suggesting that when effectively harnessed, generative AI can play a transformative role in enhancing creativity and ideation in academic environments.
Key Applications
Collaborative group-AI Brainwriting framework using LLMs
Context: Advanced undergraduate course on tangible interaction design
Implementation: Students generated ideas using a modified Brainwriting process with LLM support via Conceptboard and OpenAI's GPT-3.
Outcomes: Enhanced idea generation and evaluation, leading to a greater diversity of ideas and higher quality final projects.
Challenges: Redundancy in LLM outputs, difficulties in prompt crafting, and the need for effective integration of AI into the design process.
Implementation Barriers
Technical Barrier
Students struggled with crafting effective prompts for LLMs, impacting the quality of AI-generated ideas.
Proposed Solutions: Develop training materials for prompt engineering and create back-end interfaces for better interaction with LLMs.
Creativity Barrier
LLMs sometimes produced redundant ideas, lacking creativity.
Proposed Solutions: Investigate strategies to enhance novelty in LLM outputs, such as using diverse prompts or encouraging conceptual blending.
Project Team
Orit Shaer
Researcher
Angelora Cooper
Researcher
Osnat Mokryn
Researcher
Andrew L. Kun
Researcher
Hagit Ben Shoshan
Researcher
Contact Information
For information about the paper, please contact the authors.
Authors: Orit Shaer, Angelora Cooper, Osnat Mokryn, Andrew L. Kun, Hagit Ben Shoshan
Source Publication: View Original PaperLink opens in a new window
Project Contact: Dr. Jianhua Yang
LLM Model Version: gpt-4o-mini-2024-07-18
Analysis Provider: Openai