The dawn of intelligent creativity: Merging AI, data, and creative expressions

AI-generated legal and regulatory content should be backed by structured legal databases to avoid misinformation and legal risks. The future of AI-powered creativity will require a blend of generative capabilities with domain knowledge to ensure high-quality outputs.


CO-EDP, VisionRICO-EDP, VisionRI | Updated: 04-02-2025 16:24 IST | Created: 04-02-2025 16:24 IST
The dawn of intelligent creativity: Merging AI, data, and creative expressions
Representative Image. Credit: ChatGPT

The rapid advancements in Generative AI have fundamentally altered the landscape of creativity, automation, and knowledge management. From generating art and music to aiding in legal and scientific documentation, AI is increasingly blending with structured knowledge representations like the Semantic Web. This intersection raises important questions about the future of knowledge structuring, data transparency, and human-AI collaboration.

A recent report, titled "Semantic Web and Creative AI: A Technical Report from ISWS 2023", offers an in-depth analysis of how Large Language Models (LLMs) and Semantic Web technologies are shaping the future of AI-driven creativity and knowledge representation. The research was compiled by a team of distinguished scholars and researchers, including Stefano De Giorgis (National Research Council, Italy), Tabea Tietz (FIZ Karlsruhe, Germany), and Valentina Presutti (ISTC-CNR, Rome, Italy), among others. The study was conducted during the International Semantic Web Research School (ISWS) 2023, held in Bertinoro, Italy, and published in January 2025.

This paper explores key research questions related to the role of ontologies, knowledge graphs, and AI-generated content, addressing challenges in transparency, accuracy, and creative expression. The findings reveal both the potential and limitations of integrating Generative AI with structured knowledge representations.

Key Findings: How AI is Changing Knowledge and Creativity

The shift from traditional knowledge graphs to AI-driven models

The study explores whether Large Language Models (LLMs) will replace traditional Knowledge Graphs (KGs). While LLMs have shown remarkable success in generating text, music, and art, they often suffer from hallucination, lack of transparency, and difficulty in structured knowledge retrieval. The researchers examined whether LLMs could generate factually accurate outputs without access to structured, human-curated knowledge sources like ontologies and semantic databases.

One key finding is that LLMs still struggle with precision in structured knowledge retrieval, especially in legal and scientific domains where correctness and verifiability are crucial. The study highlights the potential for hybrid AI models, where symbolic AI (like knowledge graphs) is used alongside LLMs to enhance accuracy and explainability.

AI-Generated Music and the Role of Knowledge Graphs

A fascinating aspect of the research delves into AI-generated music and how symbolic knowledge representations can enhance creative AI systems. The study investigates whether knowledge graphs can improve AI-generated music by providing structured representations of harmony, rhythm, and historical styles.

The study used a large dataset of chord progressions and music compositions to test how well AI could generate new musical pieces using structured domain knowledge. It found that purely generative AI models struggled to integrate deep music theory concepts, whereas AI systems enhanced with knowledge graphs performed significantly better in generating harmonious compositions.

Can AI Identify Great Works of Art?

The research also investigates whether AI can distinguish between great artworks and average paintings using a combination of computer vision, knowledge graphs, and neural networks. By analyzing databases of famous artworks and their characteristics, the study attempted to train AI to recognize artistic greatness.

The findings reveal that AI can successfully identify well-known masterpieces based on structured metadata, but struggles to capture subjective elements like emotional impact, cultural significance, and artistic intent. The study suggests that while AI can assist in curating and classifying art, it cannot yet replace human judgment in determining artistic value.

AI and Legal Compliance: Is ChatGPT Providing Legally Safe Advice?

One of the most pressing issues in AI adoption is its use in legal and regulatory frameworks. The study explores whether ChatGPT and similar AI models can provide legally sound advice or whether they risk generating misleading or non-compliant information.

The research team developed a Legal Knowledge Graph (LKG) that integrates existing legal databases with AI-generated responses. This hybrid system improved the accuracy and reliability of legal AI assistants, ensuring that recommendations were factually correct and jurisdiction-specific. A key takeaway is that pure AI-generated legal content is highly risky without the backing of structured, transparent legal databases. The study calls for the integration of knowledge graphs in AI legal tools to prevent potential misinformation and legal violations.

AI Ownership and Copyright Protection in Creative Works

As AI-generated content becomes more widespread, questions of ownership and intellectual property rights are becoming critical. The study proposes a decentralized system using blockchain and semantic web technologies to track and verify AI-generated works.

The researchers introduce PICASSO, a framework designed to ensure that artists retain ownership of their AI-generated works. By combining blockchain for digital rights management and knowledge graphs for tracking artistic influence, the study offers a potential solution to AI-driven copyright disputes.

Improving AI-Generated Storytelling with Knowledge Graphs

Another area explored is how AI can enhance storytelling using structured knowledge. The researchers conducted an experiment to see whether knowledge graphs improve the coherence of AI-generated narratives.

They found that LLMs alone often generate inconsistent or logically flawed stories. However, when AI systems were enhanced with knowledge graphs, the storytelling quality improved significantly, particularly in maintaining character consistency, logical flow, and world-building.

Implications for the future: A hybrid AI approach

The ISWS 2023 study underscores that purely generative AI models are not yet sufficient to replace structured, human-curated knowledge. Instead, the future of AI lies in hybrid models that integrate structured knowledge with generative capabilities.

Some key takeaways for businesses, policymakers, and AI researchers include the importance of hybrid AI systems combining knowledge graphs with generative AI to enhance reliability in structured knowledge tasks. AI-generated legal and regulatory content should be backed by structured legal databases to avoid misinformation and legal risks. The future of AI-powered creativity will require a blend of generative capabilities with domain knowledge to ensure high-quality outputs. Additionally, blockchain and Semantic Web technologies offer a promising avenue for AI copyright and digital ownership verification. AI storytelling and content creation can also benefit from structured knowledge infusion, improving narrative consistency and depth.

  • FIRST PUBLISHED IN:
  • Devdiscourse
Give Feedback