Listening to learn: AI-ASSISTANTS 4PID concludes its collaborative actions with people with intellectual disabilities

The AI-ASSISTANTS 4PID project has reached an important milestone, concluding the collaborative actions phase that opens its development trajectory. Co-funded by the Erasmus+ Programme (Project No. 2025-1-DE02-KA220-ADU-000360510) and coordinated by IAT (Germany), the project brings together eight partners from six European countries — Germany, Italy, Portugal, Spain, Belgium, and Croatia — around a shared aim: harnessing AI-assisted solutions to enhance the independent living of persons with intellectual disabilities (PID).

The collaborative actions phase placed users at the centre of the design process. Across the six participating countries, the consortium engaged 97 participants in total: 58 people with intellectual disabilities and 39 support workers. Through facilitated sessions, the partners explored how participants approach, experience, and evaluate AI tools in real-life contexts. The resulting evidence base now serves as a roadmap for the project's next stages — the design of inclusive AI training methodologies and the development of accessible tools tailored to the priorities of the target group.

 

A confidence–utility gap that training must address

One of the most consistent findings concerns the relationship between perceived usefulness and self-confidence. Participants with intellectual disabilities readily recognised the practical benefits AI tools can offer in everyday life, from drafting messages to navigating information. Yet, when asked about their ability to use these tools independently, confidence levels remained notably low.

This gap between utility and confidence is a key insight: demonstrating AI capabilities is not enough. People with intellectual disabilities need structured, facilitated training pathways that build mastery progressively, rather than passive exposure to technology. For AI-ASSISTANTS 4PID, this finding sets a clear direction — training must be designed to close the gap, not simply to showcase what AI can do.

Anchoring AI in personal goals

A second clear pattern emerged around motivation. Activities that mirrored meaningful, everyday situations generated the highest levels of engagement. A message-writing exercise, for instance, proved markedly more motivating than technical training tasks designed to demonstrate AI features in the abstract.

This finding reinforces a principle that runs across adult education and disability research alike: learning is most effective when it is anchored in personal goals and lived experience. For the project, this translates into a clear pedagogical direction — training scenarios should be rooted in the daily aspirations of participants, from social communication to organising routines and accessing services, rather than in abstract technological features.

Support workers as co-learners, not bystanders

The collaborative actions also confirmed the decisive role played by support workers. Their competence, confidence and attitudes towards AI directly shape whether and how people with intellectual disabilities engage with the technology in real-world settings. A training methodology that only addresses end-users risks stalling at the point of adoption, because day-to-day uptake depends on the surrounding support network.

The project therefore conceives of training as a dual-track endeavour: support workers must be equipped not simply to oversee, but to co-learn alongside the people they accompany — building shared competence and a common language around AI. This dual-track approach is one of the methodological signatures that will define the next development phase.

Designing for accessibility from the outset

A further set of insights concerns the design of the AI tools themselves. Participants consistently pointed to a small number of non-negotiable design requirements: simple, uncluttered interfaces; clear step-by-step guidance; predictable interaction patterns; and robust privacy protections.

These are not stylistic preferences but accessibility prerequisites. They reaffirm what disability-led design has long argued — accessibility must be embedded in the foundation of the product, not added as a finishing touch. For AI-ASSISTANTS 4PID, this means co-designing the project's tools and resources around these principles from the earliest stages, with the target group involved in iterative feedback throughout.

Looking ahead

With this evidence base in hand, the consortium now moves into the next phase of work: translating the findings into a training methodology and a set of AI-assisted tools tailored to the realities of persons with intellectual disabilities and the professionals who support them.

The collaborative actions have provided more than data — they have shaped a clear methodological compass for what follows: training that builds confidence rather than mere exposure; tools that respond to real-life goals; and a design culture that treats accessibility as the starting point.

 

disclaimer

AI-ASSISTANTS 4PID is co-funded by the European Union under the Erasmus+ Programme (KA220-ADU). The consortium brings together IAT (Germany, coordinator), AIAS Bologna (Italy), CECD Mira Sintra (Portugal), INESC-ID (Portugal), Universitat Politècnica de València (Spain), Fundación Amica (Spain), ARFIE (Belgium), and HURT (Croatia). More information is available at ai4pid.eu.

Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or the European Education and Culture Executive Agency (EACEA). Neither the European Union nor EACEA can be held responsible for them