Skip to main content
Go to the home page of the European Commission (opens in new window)
English en
CORDIS - EU research results
CORDIS

Third wave of AI: Neuro-symbolic AI and Large Language Models

Project description

Combining LLMs and neuro-symbolic AI to revolutionise AI solutions

In recent years, advances in ΑΙ – especially large language models (LLMs) – have brought these technologies to the forefront and enabled their use across a growing number of fields. While LLMs offer numerous benefits, they also pose critical challenges, including security vulnerabilities, regulatory and ethical concerns, and the risk of hallucinations. Supported by the Marie Skłodowska-Curie Actions programme, the THIRDWAVE project is building an international, interdisciplinary network to advance LLM-driven neuro-symbolic AI. By combining the strengths of LLMs with symbolic AI, THIRDWAVE aims to create systems that are reliable, interpretable and domain-aware. These innovations will open the door to trustworthy AI applications in medicine, science, food information engineering, geodata and education.

Objective

Artificial Intelligence (AI) is transforming research, industry, and society, with Large Language Models (LLMs) playing a central role. While LLMs excel in natural language understanding, reasoning, and content generation, they also exhibit hallucinations, security vulnerabilities, ethical concerns, and regulatory issues. These challenges are particularly critical in healthcare and education, where accuracy, reliability, and fairness are essential. Addressing these shortcomings requires AI paradigms that enhance interpretability, robustness, and compliance.

THIRDWAVE aims to establish an international, interdisciplinary network to advance LLM-driven neuro-symbolic AI, integrating symbolic AI with LLMs to create interpretable, reliable, and domain-aware systems. This approach enables AI to leverage structured knowledge, improve decision-making, and comply with domain-specific constraints, making it more applicable to real-world challenges.

The project is structured around four key objectives: O1) Understanding LLMs: Analyzing capabilities and limitations to improve performance, usability, and trustworthiness. O2) Enhancing LLMs: Improving fairness, factual accuracy, and robustness through external knowledge sources and human collaboration. O3) Advancing LLM-driven Neuro-Symbolic AI: Developing hybrid systems that combine LLMs with symbolic reasoning for structured knowledge representation and better decision support. O4) Use Cases & Evaluation: Applying LLM-driven neuro-symbolic AI in healthcare, education, geodata, and food information engineering, validating scalability and societal impact.

By fostering collaboration among AI researchers, domain experts, and industry partners, THIRDWAVE will bridge the gap between data-driven and knowledge-driven AI, ensuring LLMs become interpretable, ethically aligned, and domain-aware. The project’s findings will inform AI regulation, advance research, and drive innovation, contributing to responsible AI development.

Fields of science (EuroSciVoc)

Keywords

Programme(s)

Topic(s)

Funding Scheme

HORIZON-TMA-MSCA-SE - HORIZON TMA MSCA Staff Exchanges

See all projects funded under this funding scheme

Call for proposal

(opens in new window) HORIZON-MSCA-2024-SE-01

See all projects funded under this call

Coordinator

GOTTFRIED WILHELM LEIBNIZ UNIVERSITAET HANNOVER
Net EU contribution
€ 410 820,00
Address
WELFENGARTEN 1
30167 Hannover
Germany
Region
Niedersachsen Hannover Region Hannover
Activity type
Higher or Secondary Education Establishments
Links
Total cost
No data

Participants (9)

Partners (8)