Authors:

Nachiketh Karthik

Date Created:

Not specified

Course Title:

Professor:

Not specified

About Paper:

The recent emergence of large language models (LLMs) have opened exciting possibilities for ideation and concept generation. However, understanding the true capabilities of these LLMs and utilizing them efficiently in design remains a challenge, as it requires (a) experience in using LLMs, (b) expertise in prompt engineering, (c) effective utilization of LLM outputs, and (d) comprehension of LLM functionality to explore diverse design pathways. Moreover, within the design context, an interaction pattern has been observed wherein users rely heavily on the model's responses, leading to an over-reliance and a lack of ownership over ideas. We introduce an interaction framework derived from an elicitation study that maps LLM capabilities to support design workflows in novel ways. The framework guides the development of Co-AIdeator, a system comprising a web-based interface and a task-specific design toolkit. Co-AIdeator treats the LLM as a collaborative partner, unlocking the potential for symbiotic ideation between humans and AI. By providing user-centric prompts and intuitive, context-sensitive multimodal design representations, Co-AIdeator empowers users to explore uncharted design spaces and leverage AI capabilities. This study compares the results of the Co-AIdeator system's user study with those of a baseline LLM system, drawing conclusions about our system's effectiveness in enhancing idea quality, including novelty-diversity, user engagement, and expectation alignment.

Source:

Purdue University / 2023

Topics:

No topics listed

Co-authors:

Nachiketh Karthik