At the AI Data and Robotics Forum 2024, the THEMIS project took to the stage, offering valuable insights into the evolving landscape of artificial intelligence, trustworthiness, and human-centric design.
This prestigious event, a hub for leaders, researchers, and innovators in AI, data science, and robotics, saw THEMIS participate in two influential panels addressing critical facets of AI development. In these sessions, THEMIS’s contributions underscored our mission to build trust and foster collaborative, human-centered AI systems that align with ethical values and societal needs.
The first panel on the 4th November, Standards and Metrics for Trustworthy and Safe AI, brought together experts to discuss the complex issue of establishing trust in this new technology, a challenge that goes beyond technical performance to encompass ethics, accountability, and societal impact. Representing THEMIS, Gregoris Mentzas from the National Technical University of Athens (NTUA) contributed to the discussion, sharing the project’s approach to measuring and assuring AI trustworthiness in diverse applications. As AI systems are increasingly deployed in sensitive domains such as healthcare, finance, and public administration, the question of trust becomes paramount. Trust in AI, as Gregoris explained, is not just about human confidence but also about building reliable processes within machines themselves, ensuring they can operate safely and predictably, even in dynamic environments.
Gregoris elaborated on the THEMIS project’s commitment to developing frameworks that define and measure trustworthiness in AI. He emphasized the importance of creating standardized metrics that reflect both the performance and ethical alignment of AI systems, enabling stakeholders to assess risk thresholds accurately. These metrics, Gregoris noted, must be adaptable to accommodate the rapid evolution of AI technology and the diverse contexts in which it is used. Drawing from THEMIS’s research, he discussed practical applications of these standards, such as trust metrics in data privacy and algorithmic transparency, which are designed to empower end-users and ensure that AI systems operate within agreed-upon safety margins. The goal, he explained, is to establish AI as a reliable and ethical partner in high-stakes environments, building a foundation of trust that fosters public confidence and enables further innovation.
The panel also addressed the vital role of stakeholder collaboration in setting these standards, with Gregoris underscoring the need for a multi-disciplinary approach that includes academia, industry, regulators, and civil society. By integrating diverse perspectives, the THEMIS project seeks to create a coherent framework that balances AI’s transformative potential with necessary safeguards. This collaborative spirit, as Gregoris noted, is essential for creating AI systems that not only meet regulatory standards but also resonate with societal values, ensuring that they are designed for the broader good.
In the second panel, titled From Algorithms to Assurance: Designing Human-Centric AI to Enhance Collaboration, Trust, and Acceptance THEMIS and its cluster of sister projects - TANGO Project, AI4REALNET Project, PEER AI, and HumAIne project - performed a deep dive into human-centric AI, highlighting how systems can be designed to not only work effectively with people but to actively support human decision-making, productivity, and trust.
THEMIS 5.0's Ana Correa Harcus from the Katholic University Leuven (KUL), drew from our projects research, and spoke passionately about the project’s approach to developing AI that integrates seamlessly with human needs and values whilst maintaining ethical and legal compliance. Ana presented THEMIS’s vision for human-centric AI as one rooted in empathy and respect for user autonomy. Human-centric AI, she explained, is about creating systems that are not only functional but also attuned to the psychological and social dimensions of human interaction. By understanding how people perceive, trust, and use AI, THEMIS is pioneering solutions that foster collaboration rather than control. Ana emphasized that in high-stakes applications, such as healthcare and smart cities, it is critical for AI to enhance rather than replace human expertise. This approach, she noted, aligns with the EU’s focus on AI as a tool for human empowerment, designed to amplify human potential while respecting ethical boundaries.
The workshop discussions also covered the methodological frameworks that THEMIS and its sister projects employ to design collaborative AI systems. Each project outlined how these frameworks guide the development of interfaces that prioritise transparency and ease of use, helping users build trust through clear feedback and predictable behaviour. They provided examples of their work work across sectors, from smart city planning to healthcare, where AI systems are developed to provide contextually relevant support, making them valuable partners in decision-making processes. By aligning AI with human goals, these forward-thinking initiatives aim to create a new paradigm where technology serves as a reliable ally, enhancing outcomes and fostering trust.
Another significant aspect discussed on the panel was the evaluation of trust and user acceptance in AI systems. All the projects are committed to developing robust evaluation techniques that go beyond technical performance to assess how well AI systems meet user expectations, address ethical concerns, and enhance user experience. This evaluation process is crucial for iterating and refining AI systems that truly align with human values, ensuring that they remain responsive to societal needs.
The THEMIS 5.0 representatives both enjoyed their time at ARDR24. Both panels helped to highlight THEMIS’s holistic approach to AI development, combining technical rigor with a deep understanding of ethical and social dimensions. Through discussions throughout the conference, lessons have been learned which will help THEMIS in striving to shape a future where AI serves as a responsible, trustworthy, and collaborative partner. The dual approach taken by THEIS at the conference—defining standards for trustworthiness and designing for human-centeredness—captures THEMIS’s commitment to creating AI that aligns with both regulatory requirements and the nuanced realities of human interaction.
To keep up to date with THEMIS activities and outputs please subscribe for news updates at the bottom of our home page.
Comments