Related Experiment Video
Updated: Mar 25, 2026
07:34
Perceptual and Category Processing of the Uncanny Valley Hypothesis' Dimension of Human Likeness: Some Methodological Issues
Published on: June 3, 2013
18.1K
Stable Diffusion Models Reveal a Persisting Human-AI Gap in Visual Creativity
Silvia Rondini1,2, Claudia Alvarez-Martin1,3, Paula Angermair-Barkai4,5
1Cognition and Brain Plasticity Unit, Bellvitge Biomedical Research Institute, L'Hospitalet De Llobregat, Spain.
Advanced Science (Weinheim, Baden-Wurttemberg, Germany)
|March 24, 2026
Summary
Large Language Models show limitations in visual creativity compared to humans. Increased human guidance significantly improves AI image generation, but human artists still outperform AI models.
Area of Science:
- Artificial Intelligence
- Cognitive Science
- Computer Vision
Background:
- Recent studies indicate Large Language Models (LLMs) excel in language-based creative tasks.
- Visual creativity in AI remains less explored, posing a gap in understanding AI's creative capabilities.
- Human creativity involves complex perceptual and contextual understanding, potentially challenging for AI.
Purpose of the Study:
- To compare the visual creativity of human participants (Visual Artists, Non-Artists) with an AI image generation model.
- To investigate the impact of varying levels of human input (prompting) on AI-generated image creativity.
- To evaluate AI (GPT-4o) as a rater of visual creativity and compare its judgments to human raters.
Main Methods:
- Comparative study involving human participants and an image-generation AI model.
- Two AI prompting conditions: Human-Inspired (high guidance) and Self-Guided (low guidance).
- Creativity assessment by 255 human raters and GPT-4o under two rating conditions (strict mirroring and in-context learning).
Main Results:
- A creativity hierarchy was observed: Visual Artists > Non-Artists ≥ Human-Inspired GenAI > Self-Guided GenAI.
- Higher human guidance improved AI-generated image creativity, approaching Non-Artist levels.
- Guided GPT-4o aligned better with human creativity judgments, while unguided GPT-4o showed less discrimination and inflated AI scores.
Conclusions:
- Visual creativity presents unique challenges for Generative AI (GenAI) compared to language tasks.
- Human guidance is crucial for enhancing GenAI's visual creative output.
- Distinct human capacities like perceptual nuance and contextual sensitivity may limit current GenAI visual creativity.
Related Concept Videos
01:12Modeling and Similitude
713
Scaled modeling is a fundamental technique in engineering, enabling the study of large and complex systems by creating smaller, manageable replicas that recreate critical characteristics of the original. In hydrology and civil infrastructure, for example, scaled models of dams help analyze water flow, turbulence, and pressure. This method allows for accurate predictions of real-world behavior within a controlled environment, significantly reducing the cost and time involved in full-scale...
713
01:25Creative Thinking
1.4K
Creative thinking encompasses innovative and unconventional methods for addressing challenges, often leading to groundbreaking solutions. Instead of focusing solely on enhancing existing systems, such as increasing smartphone battery capacity, creative thinking might inspire advancements like energy-efficient batteries or processors that minimize power consumption. This multidimensional approach underscores the importance of exploring novel pathways to innovation.
Divergent thinking is the...
Divergent thinking is the...
1.4K
02:45Self-Discrepancy Theory
19.0K
One influential perspective on what motivates people's behavior is detailed in Tory Higgin's self-discrepancy theory (Higgins, 1987). He proposed that people hold disagreeing internal representations of themselves that lead to different emotional states.
19.0K
02:16Stereotype Content Model
15.7K
The Stereotype Content Model (SCM) was first proposed by Susan Fiske and her colleagues (Fiske, Cuddy, Glick & Xu, 2002; see also Fiske, 2012 and Fiske, 2017). The SCM specifies that when someone encounters a new group, they will stereotype them based on two metrics: warmth—or that group’s perceived intent, and how likely they are to provide help or inflict harm—and competence—or their ability to carry out that objective. Depending on the warmth-competence...
15.7K
01:16Non-equilibrium in the Cell
5.6K
An important concept in studying metabolism and energy is that of chemical equilibrium. Most chemical reactions are reversible. They can proceed in both directions, releasing energy into their environment in one direction, and absorbing it from the environment in the other direction. The same is true for the chemical reactions involved in cell metabolism, such as the breaking down and building up of proteins into and from individual amino acids, respectively. Reactants within a closed system...
5.6K
01:21Cognitive Learning
1.5K
Cognitive learning is based on purposive behavior, incidental learning, and insight learning.
E. C. Tolman's theory of purposive behavior emphasizes that much behavior is goal-directed. He argued that to understand behavior, we must look at the entire sequence of actions leading to a goal. For instance, high school students study hard, not just due to past reinforcement but also to achieve the goal of getting into a good college.
Tolman introduced the idea that behavior is influenced by...
E. C. Tolman's theory of purposive behavior emphasizes that much behavior is goal-directed. He argued that to understand behavior, we must look at the entire sequence of actions leading to a goal. For instance, high school students study hard, not just due to past reinforcement but also to achieve the goal of getting into a good college.
Tolman introduced the idea that behavior is influenced by...
1.5K