The AIXPERT project proposes a novel, comprehensive approach to developing explainable, accountable, and transparent AI systems. At its core, the project introduces an adaptable, situation-aware AI-agentic platform capable of encapsulating various AI models, regardless of their underlying architecture. This architecture-agnostic approach significantly enhances the trustworthiness of AI systems by providing a consistent framework for explainability and accountability across different model types. The project aims to address challenges in AI explainability, transparency, accountability, autonomy, and robustness by integrating multi-agent systems with multimodal foundation models and leveraging real-time human feedback. This integration enhances AI system trustworthiness and user-friendliness while maintaining flexibility in the choice of underlying AI models.
The project will be implemented in a multi-layered architecture, with each layer focusing on specific aspects of AI interpretability and explainability: the Agent-World Interface Layer will define and coordinate AI agents, their situational awareness, and their interactions with real-world knowledge sources; the Dialogue Mediation Layer will control user-agent and agent-agent communications; and the Cognitive Foundation Layer will provide the base capabilities for the system based on explainable multimodal foundation models. The project will also focus on developing a framework for assessing AI trustworthiness, demonstrating the value of explainable AI through pilot demonstrations in healthcare, recruitment services, manufacturing, educational robotics, and creative industries, and ensuring the sustainability of its results.
AIXPERT envisions delivering AI solutions that are not only transparent and ethical but also sustainable and adaptable to diverse user needs, ultimately fostering greater trust in AI across multiple industries and societal sectors.
---
AIXPERT project has received funding from the European Union’s Horizon Europe research and innovation programme under Grant Agreement No 101214389. This work has received funding from the Swiss State Secretariat for Education, Research and Innovation (SERI).