In this episode of the ELOQUENCE podcast series, the focus was on the piloting phase of the project and the role of human-computer interaction in testing AI-driven language technologies in real-world scenarios.
The guest was Dr. Giuseppe Caggianese, Research Scientist at the National Research Council of Italy and leader of Work Package 5 in the ELOQUENCE project. With expertise in human-computer interaction, adaptive interfaces, virtual and augmented reality, and AI usability, Jeppe offered insight into how intelligent systems can be designed, tested, and improved with users’ needs at the centre.
Jeppe began by reflecting on his journey into artificial intelligence and human-computer interaction. His research career started during his PhD, where he worked with AI and GPU parallel computation to simulate the behaviour of agents in digital environments. This work led him toward virtual and augmented reality, and eventually to a deeper interest in how people interact with technology. One aspect that particularly shaped his research was the observation that users often interact with interfaces in unexpected ways. For designers and developers, this highlights the importance of involving real users in the design process from the beginning.
A central part of the conversation focused on what piloting means within the ELOQUENCE project. In this context, piloting refers to implementing and validating the technologies developed across the project in concrete scenarios. ELOQUENCE will test its tools through four pilots covering smart home, social, and safety-critical contact centre contexts. These pilots will also involve multiple languages, including Spanish, Italian, Greek, Serbian, and English, making multilingual testing a key part of the validation process.
Jeppe explained that the project follows an iterative, user-centred methodology. This means that technologies are not only tested for technical performance, but also for feasibility, usability, applicability, and their impact on user experience. Feedback collected from users will help optimise the AI components and improve how the systems perform in different contexts.
The episode also addressed some of the challenges involved in piloting AI technologies. When systems interact directly with people and may influence their choices, testing must be careful and responsible. This is especially important when working with new AI tools that may still be immature or require some level of control before being made widely available. ELOQUENCE aims to address these challenges by developing technologies that are aware of ethical and social requirements, able to consider context, reduce hallucinations, mitigate biased answers, and provide more trustworthy outputs.
Another important topic was the role of human factors in AI interaction. In safety-critical contexts, users may interact with AI under stress, while in social scenarios, cultural and linguistic differences can shape how people understand and trust the system. This makes adaptive, culturally sensitive, and easy-to-use interfaces essential.
Looking ahead, Jeppe pointed to a future where AI is no longer seen only as a standalone tool, but as a collaborative partner that can augment human capabilities. For this to work, AI systems must be transparent, accountable, usable, and designed around real human needs.
Listen to the full episode here.
