Enterprise AI Analysis
Are Semantic Networks Associated with Idea Originality in Artificial Creativity? A Comparison with Human Agents
Umberto Domanti, Lorenzo Campidelli, Sergio Agnoli, Antonella De Angeli
The application of generative artificial intelligence in Creativity Support Tools (CSTs) presents the challenge of interfacing two black boxes: the user's mind and the machine engine. According to Artificial Cognition, this challenge involves theories, methods, and constructs developed to study human creativity. Consistently, the paper investigated the relationship between semantic networks organisation and idea originality in Large Language Models. Data was collected by administering a set of standardised tests to ChatGPT-40 and 81 psychology students, divided into higher and lower creative individuals. The expected relationship was confirmed in the comparison between ChatGPT-40 and higher creative humans. However, despite having a more rigid network, ChatGPT-40 emerged as more original than lower creative humans. We attributed this difference to human motivational processes and model hyperparameters, advancing a research agenda for the study of artificial creativity. In conclusion, we illustrate the potential of this construct for designing and evaluating CSTs.
Executive Impact: AI in Creativity
This study provides critical insights into the performance and underlying mechanisms of AI in creative tasks compared to human agents. Understanding these differences is key for designing effective AI-powered creativity support tools.
0
ChatGPT-40 Mean Fluency
0
HCH Peak Originality Score
0
GPT-40 Peak Originality Score
0
GPT-40 Originality Over LCH
Deep Analysis & Enterprise Applications
Select a topic to dive deeper, then explore the specific findings from the research, rebuilt as interactive, enterprise-focused modules.
Research Context
Methodology
Key Findings
Enterprise Implications
Case Study: AI's Creative Potential & Copyright
A provocative example is Théâtre D'opéra Spatial, a "painting" generated by Midjourney following a prompt by Jason M. Allen. The artwork won the fine art competition at the 2022 Colorado State Fair. Yet, the U.S. Registration Office denied copyright to the artist based on the impossibility of differentiating between human and artificial contributions in the creative process [20]. This highlights emerging challenges in ownership and authenticity with AI-generated content.
Creative Cognition Framework
Memory
→
Attention
→
Control
→
Creative Thought
The creative process, as defined in Artificial Cognition, emerges from basic neurocognitive abilities (Memory, Attention, Control) which then converge into generative modalities such as divergent and convergent thinking.
| Paper | Comparison | Creativity Assessment | Main Limitations | Main Findings |
|---|---|---|---|---|
| [108] | 42 psychology students VS ChatGPT-3 on AUT. |
|
Small human sample. |
|
| [11] | 20 psychology students in cou-ples vs 10 chatbots on AUT. |
|
Small human sample. |
|
| [65] | 100 participants vs 6 chatbots (Alpa.ai, Copy.ai, ChatGPT-3/4, Studio.ai, YouChat) on AUT. | Six human coders and OCSAI rated originality. | Paid Prolific participants. | LLMs = H. 9% of H > LLMs. |
| [45] | Large human baseline [83] vs ChatGPT-3/4 on DAT. | Semantic distance. | Small differences in mean scores between groups. | LLMs > H; Concerns on AI reliabil-ity. |
| Analysis Measure | Acronym | Description | Expected Results |
|---|---|---|---|
| Average Short Path Length | ASPL | Average shortest number of steps (i.e., edges) needed to traverse between any pair of nodes. A higher value means the network is more spread out. | LCH > GPT > HCH |
| Connectivity | CC | Clustering coefficient of a network, describing how much the neigh-bours of a node are interconnected. A higher value means more interconnection. | LCH < GPT < HCH |
| Modularity | Q | Degree to which a network can be divided into sub-communities. A higher value means more segregation. | LCH > GPT > HCH |
| Percolation Resiliency | R | Measure of flexibility of a network. A higher value means more resilience. | LCH < GPT < HCH |
0
GPT's ASPL is 1.05 points higher than HCH, indicating a more spread-out network
The analysis confirmed that ChatGPT-40 exhibits a less flexible semantic network compared to higher creative humans, as indicated by a significantly higher Average Short Path Length (ASPL) of 4.378 for GPT vs. 3.325 for HCH.
| Hypothesis | Expected direction | Result |
|---|---|---|
| Idea Originality | HCH > GPT > LCH | HCH > GPT > LCH |
| Network Flexibility | HCH > GPT > LCH | HCH > LCH > GPT |
| Measure | Comparison | df | t-statistic | p-value | d | Direction |
|---|---|---|---|---|---|---|
| ASPL | GPT vs LCH | 1998 | 51.91 | < .001 | 2.32 | GPT > LCH |
| ASPL | GPT vs HCH | 1998 | 56.93 | < .001 | 2.55 | GPT > HCH |
| CC | GPT vs LCH | 1998 | -8.76 | < .001 | 0.39 | GPT < LCH |
| CC | GPT vs HCH | 1998 | -19.84 | < .001 | 0.89 | GPT < HCH |
| Q | GPT vs LCH | 1998 | 31.42 | < .001 | 1.41 | GPT > LCH |
| Q | GPT vs HCH | 1998 | 37.17 | < .001 | 1.66 | GPT > HCH |
Operational Insights & Methodological Recommendations
The study provides key insights for developing Creativity Support Tools (CSTs) and advancing the field of artificial creativity:
OPI1 Artificial Creativity is Average Performance: ChatGPT-40's mean originality scores reflected average human performance, excelling over lower creative individuals but not reaching the highest creative human levels. This suggests AI is beneficial for users seeking improved novelty, but designers must ensure new interfaces bridge human-AI creativity effectively to leverage its average nature for broader user benefit.
OPI2 Artificial Creativity is a Process, not only a Product: The findings provide preliminary evidence of a partial association between idea originality and semantic networks in ChatGPT-40. Despite a rigid network, AI's hyperparameters may partly compensate for structural limitations, enabling original outputs. Future research should focus on supporting different stages of the creative process and understanding how AI affects user performance.
MER1 Human Creativity Provides Methodological Scaffolding: Artificial Cognition draws parallels between AI and the human mind, highlighting differences in originality and semantic memory. The creative cognition framework (Memory, Attention, Control) provides concepts for studying both human and artificial creativity, addressing micro-level connections between machine hyperparameters and human executive functions.
MER2 Collecting Reliable Human Baselines: The human sample is critical. Creativity varies across individuals and lifespan, influenced by cognitive functions, motivations, and experiences. A median split (LCH/HCH) offers finer-grained operationalization, underscoring the need for careful consideration of individual and contextual factors in experimental design.
MER3 Artificial Creativity Provides Metaphors for Design: The semantic network metaphor offers stimulating design trajectories for HCI research. New CSTs could allow users (artists, researchers) to visualize and manipulate semantic networks, exploring new semantic spaces and fostering creative outcomes. This requires diverse methods for testing humans and machines, extending beyond prompt engineering to user research.
Calculate Your Potential AI ROI
Estimate the financial and efficiency gains your enterprise could realize by strategically integrating AI solutions.
Estimated Annual Savings
$0
Annual Hours Reclaimed
0
Your AI Implementation Roadmap
Our structured approach ensures a smooth, impactful AI integration tailored to your enterprise's unique needs and objectives.
Phase 01: Discovery & Strategy
Comprehensive analysis of existing workflows, identification of AI opportunities, and strategic planning aligned with business goals. Define KPIs and success metrics.
Phase 02: Pilot & Proof-of-Concept
Develop and deploy a small-scale AI solution to validate its efficacy, gather initial feedback, and demonstrate tangible value within a controlled environment.
Phase 03: Scaled Implementation
Roll out the validated AI solution across relevant departments, ensuring seamless integration with existing systems and robust performance monitoring.
Phase 04: Optimization & Future-Proofing
Continuous monitoring, performance tuning, and iterative improvements. Explore advanced AI capabilities and expand the solution's scope for long-term growth.
Ready to Transform Your Enterprise with AI?
Our experts are ready to guide you through the complexities of AI integration, from strategic planning to implementation and optimization.
Ready to Get Started?
Book Your Free Consultation.
Let's Discuss Your AI Strategy!
Lets Discuss Your Needs
Select a Date
Sun
Mon
Tue
Wed
Thu
Fri
Sat