Recent advances in deep generative neural networks have made it possible for artificial intelligence to actively collaborate with human beings in co-creating novel content. In a study published in AC Digital Library, the researchers analyzed the 'Uncovering the Roles of Deep Generative AI during Social Music Composition'.  

The research was a quantitative lab study. The researchers observed 30 participants compose a musical phrase in two pairs, both with and without AI. The study reveals that AI may play essential roles in influencing human social dynamics during creativity, including: 

• Implicitly seeding a common ground at the start of a collaboration 
• Acting as a psychological safety net in creative risk-taking,  
• Providing a force for group progress,  
• Mitigating interpersonal stalling and friction, and  
• Altering users' collaborative and creative roles. 

The research investigates how pairs of people collaboratively create artistic content with deep generative AI and how social dynamics can be affected by the presence of AI. The study's findings lay the groundwork for understanding the future of generative AI in social creativity, suggesting implications for how AI could enrich, impede, or alter creative social dynamics in the years to come.  

Human-AI co-creation 

According to the report, the study provides a foundational perspective on human-AI co-creation in contexts where multiple people co-create together, illuminating how AI may alter the interpersonal dynamics within a team. While integrating AI into the human workflow has always been an essential area of research, there is increasing value in introducing Computer-Supported Cooperative Work (CSCW) perspectives in Human-AI collaboration research.  

Prior work on collaborative technologies for music-making has examined how users use tools and systems while co-composing or co-curating music. Here, to understand the potential for AI to assist with co-creation, the researchers conducted a qualitative laboratory study using a previously developed interface to a deep generative AI, Cococo.  

Cococo is a web-based music editor for human-AI co-creation that augments standard generative music interfaces with a set of AI steering tools. Each participant independently completed an online tutorial on the interfaces and composed two pieces. Later participants compared and contrasted their experiences co-composing with and without the model via a semi-structured interview. 

Analyzing the result 

Though participants' co-creation process with AI present was largely similar to their process without AI, they had additional decision-making points on when, where, and how to use AI in their collaboration. These additional decisions included deciding where and when to request that the AI generate content, what quality of music they wanted in that segment, evaluating and selecting which of the N AI-generated options sounded best and deciding which partner would be the main driver of these decisions above, each time they occurred.  

The study denotes that AI establishes a common ground, plays the role of a psychological safety net and is a force for progress in social collaboration. It also acts as a social lubricant and a force that alters the creative and collaborative roles of humans. 

Conclusion 

The psychological safety net provided by AI allowed users to feel more comfortable taking creative risks with a partner watching. The study proposes that future systems and studies could investigate how to leverage the playful properties of AI as a psychological safety net, particularly in "icebreaker" phases of collaboration where collaborators have yet to build up an adequate level of mutual trust or when creative risk-taking is detected to be low. 

Although the positive implications of AI are evident here, there are some limitations that should be dealt with. First, some level of creative tension is likely to arise across a wide range of tasks. While the study found that AI served as a mitigator of interpersonal friction, the nuances of those dynamics could be affected by the specific context and nature of the task. Thus, future research should further examine the role of AI across a greater variety of human-human collaboration contexts (e.g., different tasks and social relationships among human collaborators). 

Image source: Unsplash

Content source: AC digital library