During the first 15 months (M1–M15), the SKillAIbility consortium executed all planned technical and scientific activities, with no deviations. Achievements are presented below according to the project’s four objective categories.
Individual objectives – human‑centric assessment and foresight. A systematic state‑of‑the‑art review of AI/automation in manufacturing was completed, covering workforce transformation, digital inclusion and policy readiness (D2.1). Four megatrend families were analysed using OECD, WEF and JRC foresight methods, producing a trend radar and a non‑linear organisational adaptability framework. Two co‑creation workshops (50 participants) defined skills and human task needs, resulting in a Deaf and Hard of Hearing (DHOH) needs questionnaire translated into five languages. The overarching SKillAIbility framework was designed, including ontologies for skills (ESCO), technologies, and human factors (physical, cognitive, perceptual, psychosocial). Four pathways – augmentation, empowerment, inclusivity, symbiosis – were defined and published with a DOI (10.5281/zenodo.18234137). Human‑centric criteria for AI/automation complementarity were defined based on the four human factor dimensions. Two experimental campaigns tested collaborative robot assembly and AI‑based predictive maintenance with 20 participants (including deaf workers), showing that shared control reduced error rates for novice workers and that deaf participants required visual alerts and captioning. A persona‑based modelling approach was designed for five worker profiles. Two master theses produced a five‑level maturity model for augmentation and inclusivity and a design framework for human‑machine symbiosis (preserve human agency, enable reciprocal learning, design for emergent outcomes).
Technical objectives – AI and automation design. A systematic literature review synthesised seven design principles for human‑centric AI (e.g. preserve human agency, design for calibrated trust, bring users into early problem framing). The echelonised Design Science Research (eDSR) methodology was deployed across four use cases (gesture control, AR‑guided assembly, LLM‑based tutor, 3D printing learning factory), defining validation criteria for each echelon. Two automation experiments showed that advisory configuration achieved highest trust and that participants over‑relied on automation in supervisory mode.
Economic objectives – work organisation and investment incentives. A Human Centric Learning Factory methodology was developed, integrating Bloom’s taxonomy and an Actor‑Role‑Skill taxonomy, covering human centricity, digital technologies, training modalities, and accessibility adaptations for deaf, low‑skilled and ageing learners. A stakeholder survey (85 responses) found digital/AI skills shortages, training‑industry misalignment, and preference for financial incentives. A policy gap analysis identified insufficient policies for AI‑human collaboration skills and Learning Factory recognition. A pre‑standardisation roadmap was drafted, mapping relevant standards (ISO 9241, ISO 13482, CEN/CLC/JTC 21) and identifying gaps.
Social objectives – training methodologies for upskilling. Gaps in VET provision were identified (lack of human‑AI collaboration modules, inaccessible digital tools, insufficient inclusive pedagogy). Four use cases were fully defined (empowerment, symbiosis, augmentation, inclusivity). A quantitative survey instrument (company and employee level) was drafted, targeting firms across five EU countries using difference‑in‑differences and propensity score matching (ethics approval pending).
