Perception of AI and Human Collaboration
Description
Research Hypothesis The study hypothesizes that design students' perceptions of authorship in AI-assisted design vary depending on the degree of influence their original image prompts have on the final AI-generated outcomes. It aims to understand at what point students consider AI-generated images their own creation and how comfortable they are in collaborating with AI tools, giving credit to the AI where due. Data Collection &Analysis Data was gathered through a mixed-methods approach involving an AI-aided design workshop followed by a survey. Participants included second-year industrial design (ID) and interior architecture (IA) students. The workshop introduced the Vizcom AI tool, which students used to generate AI-assisted designs based on their studio project outcomes. The survey assessed students' familiarity with AI tools, perceptions of authorship, reasons for their perceptions, and comfort levels in co-designing with AI. The sample size was 30 students (12 ID, 18 IA). difference in familiarity level with AI tools (Q1), perception of authorship (Q2), reasons of perception of the puthorship (Q3), and comfort level of collaboration with ai (Q4) between ID and IA students. A Spearman's rank correlation coefficient can be calculated to assess the relationship between students' familiarity with AI tools (Q1) and their comfort levels in co-designing with AI (Q4). Textual analysis can be conducted to thematically analyze the responses from the open-ended question, Q5. Notable Findings T-Test: The T-test results indicate that there is no statistically significant difference between ID and IA students in terms of their familiarity with AI tools, perception of authorship, reasons for their perception of authorship, and comfort level with collaborating with AI. Perception of Authorship: 33.3% of students considered AI-generated images as their own creation up to a 50% image prompt influence level. Significant thresholds for authorship were identified between 40% and 70% image prompt influence, with most students feeling a sense of ownership within this range. Comfort with AI Collaboration: 46.4% of participants were comfortable or very comfortable collaborating with AI and might give credit to AI. A moderate positive correlation (rs=0.48) was found between familiarity with AI tools and comfort in co-designing with AI. Reasons for Perceived Authorship: The primary reason cited was that the design was still based on the students' original research or findings (60%). Implications for Future Research The study's findings suggest a need for further research with a larger and more diverse sample. Exploring perceptions of more experienced designers and different design fields could provide a broader understanding of authorship in AI-assisted design. Future studies could also delve deeper into the reasons behind neutral attitudes towards AI collaboration and explore the impact of storytelling in enhancing AI-human collaboration.
Files
Steps to reproduce
Study Design -The study used a mixed-methods approach, combining an AI-aided design workshop with a survey to collect both quantitative and qualitative data. Participants -The study involved 30 second-year students: 12 from Industrial Design (ID) and 18 from Interior Architecture (IA). Workshop Protocol -The 50-minute workshop introduced students to the Vizcom AI tool. -First 20 minutes: Introduction to the tool's features and use. -Next 20 minutes: Practice session where students used their project outcomes as prompts to generate AI-assisted designs. They created 11 images by adjusting the image prompt influence from 100% to 0% in 10% decrements. Suggested prompts were "modern design of a salt and pepper grinder" for ID and "a modern and sleek house exterior/interior" for IA. Survey Protocol -After the workshop, students completed a 10-minute survey. Survey Questions: -Familiarity with AI tools (Likert scale: 1-5). -Perception of authorship at different AI contribution levels (percentage-based scale: 0%-100%). -Reasons for authorship perception (categorical with open-ended option). -Comfort in co-designing with AI and willingness to give credit (Likert scale: 1-5). -Open-ended question on overall thoughts about using Vizcom AI. Instruments and Software -Vizcom AI Tool: Used in the workshop. -Survey Software: For administering and collecting responses. -Statistical Software: Excel/SPSS/R for t-tests and correlation analysis. -Qualitative Analysis Software: Qualtrics for thematic analysis. Data Analysis Workflow Quantitative Data Analysis -Descriptive statistics summarized frequencies and percentages. -Independent samples t-tests compared ID and IA student responses. -Spearman's rank correlation coefficient assessed the relationship between AI familiarity and comfort in co-designing. Qualitative Data Analysis -Thematic analysis of open-ended responses involved coding, categorizing, and interpreting themes. Reproducibility To reproduce the study: -Recruit second-year ID and IA students. -Conduct a 50-minute workshop introducing the Vizcom AI tool. -Administer a post-workshop survey assessing AI familiarity, authorship perceptions, comfort levels in co-designing with AI, and overall thoughts. -Use electronic survey platforms and ensure data anonymity. -Analyze data with statistical and qualitative software, following the outlined workflows. -By following these protocols, researchers can replicate the study to explore design students' perceptions of authorship in AI-assisted design.