Should the development of AI be accompanied by an understanding of its potential impact on people’s everyday lives and decision-making processes? We believe it should. Our artificial intelligence research allows us to examine technology from the user’s perspective, recognising it as both an innovative tool and an integral part of professional and social life.
We collaborate with universities, institutions and technology partners on research projects that combine data analysis with empathetic experience design. Our approach is centred around questions concerning the clarity, usability, trustworthiness and ethics of AI-based solutions.
AI research should be conducted in collaboration with the people for whom this technology is being created. Find out more about our projects, in which we acted as a research partner.
AI chatbot powered by GDPR and patent rulings
In one of our research projects, conducted in partnership with a consortium comprising the AGH University of Science and Technology and the University of Luxembourg, we had the opportunity to examine a prototype AI chatbot designed to assist users in interpreting GDPR regulations and patent rulings. This solution aimed to make specialist legal knowledge accessible to people who do not have regular access to a lawyer at work, in a user-friendly form.
Our task was to examine how users perceive the chatbot’s communication style and the extent to which they find the AI-generated answers understandable, trustworthy and useful. To this end, we designed and conducted remote usability tests and one-to-one interviews with entrepreneurs, inventors, and company specialists.
One of the participants was a physiotherapist who was planning to patent an innovative solution for his practice. Thanks to the chatbot, he was able to obtain preliminary information about patent procedures and intellectual property protection independently, which would normally require time-consuming legal consultations.
This illustrates the potential of such tools for micro-entrepreneurs and self-employed specialists.
The effectiveness of AI solutions hinges not only on the quality of the algorithm, but also on how technology interacts with people.
AI Assistant in the Doctor’s Office
A doctor’s daily work requires medical knowledge, effective communication, quick decision-making and the simultaneous documentation of visits. In a research project conducted in partnership with a technical university and a company from the medical sector, we explored how AI-based technology could support these processes without disrupting the natural rhythm of a doctor’s work.
The starting point was the idea of a voice-based AI assistant that discreetly records the course of the consultation and automatically generates notes, thereby reducing the doctor’s administrative workload.
Even at the planning stage, we asked ourselves a fundamental question: what can be digitised without interfering with the doctor’s workflow? The key was to set a clear boundary: the system was intended to provide support, but not to suggest diagnoses or therapies.
Next, as part of the project, we moved into the research phase, conducting individual interviews with doctors and carrying out quantitative surveys to assess the impact of a real AI solution on the course of a visit and the doctor–patient relationship.
Our goal was to capture both the functional aspects of the tool and the expectations of the specialists who would use it. The research focused on answering questions such as:
- In what ways could the proposed voice-based medical documentation system influence the course of a medical visit?
- What potential scenarios exist for using a voice assistant to support doctors in completing documentation?
- How do doctors evaluate the usefulness and usability of the solution being developed?
- What opportunities and challenges are involved in implementing such a solution?
- Whether there are any technical barriers (including doctors’ digital skills) that could affect the tool’s effectiveness?
From the doctors’ perspective, two issues emerged as crucial: the transparency of AI’s operations and its ‘invisibility’ during the visit. Many respondents emphasised that technology can be helpful as long as it does not interfere with the conversation, dominate the meeting or impose an unnatural course of action.
During the project, we also faced questions that went beyond the mere functionality of the tool:
- when implementing AI, are we truly addressing users’ needs, or are we assuming that we know better what they require?
- should the mere existence of a technology be enough to justify its implementation? Or should we consider whether it might actually disrupt natural processes rather than making work easier?
These are questions of ethics and responsibility, and of designing the future in a mindful way. In our research, we not only test interfaces and functions, but also recognise whether AI is appropriate in a given context and who is responsible for the consequences of its operation. A hybrid approach combining technological capabilities with a genuine, empathetic understanding of user needs is especially essential in healthcare.
In the context of background systems, the question of designer and developer responsibility is particularly important: the less visible a tool’s operation is, the greater the responsibility of those who create and implement it.
Users may not understand the mechanisms behind a generated response, which is why we must ensure that they align with users’ needs and not just the capabilities of the technology.
AI in the organization: an internal knowledge chatbot for employees
In a project carried out with Philip Morris International, we explored the implementation of an AI chatbot to support knowledge management in a corporate environment. The aim of the solution was to enable employees to quickly find answers to questions about procedures, tools and internal processes, such as ‘What should I do if I change departments?’ and ‘How can I transfer to another location?’.
Our task was to examine how employees respond to this type of tool, considering whether they find it useful, understandable and trustworthy, and what their concerns and expectations are regarding this form of support.
The exploratory project was based on experimental testing and elements of the scenario method. Participants were asked to immerse themselves in situations they might encounter in the workplace, such as changing departments or relocating to another office, and then ask the chatbot the kinds of questions they would naturally ask in such situations. We observed their reactions not only to the content of the responses, but also how they formulated questions, their doubts and their level of trust in the information provided.
One of the most interesting aspects of the project was analysing the conditions under which AI can genuinely simplify processes, and when it becomes an unnecessary layer. During the research, we found that the tool performed well when answering procedural and standard questions. However, in more complex situations, a greater level of contextual and nuanced alignment was expected. This is less a matter of the technology itself and more a matter of understanding in which areas it is genuinely worth applying.
From a research perspective, we also ask ourselves whether implementing AI in such cases is worthwhile. Is the technology a response to a genuine user need? Should we deploy it simply because we have access to it, or only when it adds genuine value without blurring lines of responsibility?
When testing AI in organisations, it is important to consider not only the quality of the technology’s responses, but also the context in which they are received: who is asking, in what situation, and what emotions and expectations are involved. This is because, in the workplace, trust in information generated by a machine still depends on whether the user knows who is behind it.
Human at the center
Although the projects described covered different topics and environments, they all had one thing in common: the human being was the point of reference, not the technology. At EDISONDA, our approach to AI is not solely focused on performance, efficiency or innovation. We look deeper. We ask whether the solution is understandable, useful, necessary and responsible.
In our view, AI research encompasses more than just data analysis and algorithm testing. Above all, it is about listening to the people for whom these tools are being created. We believe that technology can only be developed in this way — technology that works, inspires trust, supports everyday tasks and respects the boundaries of relationships, trust and autonomy.
What connects our approach across all projects is attentiveness: to context, to the user and to consequences. This enables us to improve solutions and recommend effective implementation strategies that are well received by users and address their sometimes unspoken problems and needs.
