International Research Week on AI and Social Services

Situated within one of the world’s most highly digitalized public sectors, Danish social work has been shaped by extensive investments in digital infrastructure and data-driven governance. This keynote explores the development of digital technologies and particularly AI in Danish social work over the last decade. The presentation examines how these developments have intersected with a field characterized by individualized, case-oriented interventions and a strong emphasis on statutory practice.

The keynote traces three waves of digitalization: first, the introduction of digital case management systems and increasing standardization of social work; second, the growing use of data collection, performance metrics, and predictive algorithms; and third, the recent emergence of generative AI and automated text production. Drawing particularly on the Danish Public Employment Services (PES) as a case, the presentation discusses examples such as the STAR predictive algorithm and recent applications of generative AI, including Fasit.

Overall, the keynote aims to reflect on how AI is transforming professional practice, organizational control, and decision-making in contemporary Danish social work.

Situated within one of the world’s most highly digitalized public sectors, Danish social work has been shaped by extensive investments in digital infrastructure and data-driven governance. This keynote explores the development of digital technologies and particularly AI in Danish social work over the last decade. The presentation examines how these developments have intersected with a field characterized by individualized, case-oriented interventions and a strong emphasis on statutory practice.

The keynote traces three waves of digitalization: first, the introduction of digital case management systems and increasing standardization of social work; second, the growing use of data collection, performance metrics, and predictive algorithms; and third, the recent emergence of generative AI and automated text production. Drawing particularly on the Danish Public Employment Services (PES) as a case, the presentation discusses examples such as the STAR predictive algorithm and recent applications of generative AI, including Fasit.

Overall, the keynote aims to reflect on how AI is transforming professional practice, organizational control, and decision-making in contemporary Danish social work.

Artificial intelligence offers significant opportunities to enhance participation for people with disabilities. AI-based technologies can make communication and information more accessible, support different forms of expression, and reduce barriers in digital environments. At the same time, the increasing integration of AI into communication and social services can also create new barriers or reproduce normative assumptions about communication, ability and appropriate behaviour.

Drawing on findings and experiences from several research and development projects, the presentation looks at this ambivalence between enabling participation and producing new forms of digital exclusion. A special focus is put on the DiBaKo project, in which people with different disabilities tested an AI-supported system integrating speech recognition, speech synthesis and emotion recognition into videoconferencing. Similar questions are currently further investigated in the DFG-funded research initiative SPEECHES, where the Multi-Modal Online Counseling project deals with AI-supported transcription, content analysis, emotion recognition and language simplification in psychosocial online counseling.

The findings from DiBaKo show that the effects of AI-based assistance cannot be understood only from the technical functionality. Speech recognition may reduce cognitive barriers while at the same time remaining inaccessible to blind users; speech synthesis can strengthen communicative agency, but also raises questions about voice, identity and control; and emotion recognition may provide orientation while also introducing normative interpretations of emotional expression. An important point is that many of these tensions and design requirements only became visible through the active involvement of users themselves.

The presentation therefore argues that participatory development is an important condition for inclusive AI. People with disabilities and users of social services should not only be addressed as recipients or test users of technological solutions, but should be involved as experts in defining problems, evaluating technologies and shaping their further development. The experiences from DiBaKo and the ongoing work in SPEECHES show the necessity to understand inclusive AI as an ongoing negotiation between technological possibilities, individual needs, professional contexts and user perspectives. Participation is therefore not only an additional feature of inclusive AI development, but a prerequisite for making sure that AI expands opportunities for participation instead of creating new forms of exclusion.

Two major transformations are currently shaping social work, health and education: the rapid development of digital technologies and artificial intelligence, and the growing urgency of sustainability and climate-related challenges. Although often discussed separately, these developments increasingly influence one another in professional practice.

This presentation explores a central question: Under what conditions can AI contribute to more sustainable social work? It considers social, ecological and economic sustainability, examining potential benefits such as improved access to services and reduced administrative workloads alongside risks to inclusion, professional relationships and long-term organisational autonomy.

Drawing on a conceptual framework developed in the author’s book chapter manuscript, the presentation introduces a reflection matrix that connects three areas of digitalisation—professional activities, tools and professional discourses—with the three sustainability dimensions (Siemer, n.d.). An illustrative case of an AI-supported counselling system demonstrates how the matrix can make competing priorities visible and support reasoned professional decisions. The matrix is presented as a reflective tool rather than a validated assessment instrument.

Recent literature on AI’s environmental impacts broadens this analysis beyond electricity consumption to carbon emissions, water use, material requirements and the unequal geographical distribution of environmental burdens (Gröger et al., 2025; Aczel et al., 2026; Barnett-Itzhaki, 2026). These concerns are considered alongside ethical questions of fairness, accountability and professional responsibility in social work (Reamer, 2023).

The presentation examines the relationship in both directions: how AI may support or undermine sustainability, and how sustainability goals can guide its development, selection and use. It offers a practical basis for discussion and for critical professional judgement about socially and ecologically responsible digitalisation.

Artificial intelligence is increasingly influencing professional practice in social work. While discussions about AI often focus on technological developments or ethical concerns, there is still limited clarity regarding the specific competencies social work professionals need in order to work critically, responsibly and effectively with AI-supported systems.

Several competency frameworks already exist, also specifically for AI. However, these models are often developed within sector-specific contexts and may not fully address the relational, ethical and justice-oriented dimensions that characterize social work practice.

This interactive session aims to explore how existing (AI) competency models can inform the development of a competence framework specifically for social work professionals. Drawing on emerging work, including recent initiatives around AI competencies for nurses, participants will collaboratively reflect on which knowledge, skills, attitudes and ethical capacities are needed in social work contexts.

The session will invite participants to compare international perspectives, identify overlaps and gaps between existing models, and discuss how AI competencies can be embedded within professional values such as participation, inclusion, empowerment and social justice. Through dialogue and collaborative exchange, the workshop seeks to contribute to the development of a shared and practice-oriented understanding of AI literacy and professional competence in social work.

Assessing distress in adults with profound autism and minimal or absent functional speech relies largely on behavioral observation, which may not capture early changes preceding behavioral escalation. This study aims to identify early physiological and visual-attentional indicators of distress and examine their correspondence across a controlled virtual reality (VR) environment and everyday activities.

A five-week observational study is planned with ten adult men aged 25–35 years with autism, co-occurring intellectual disability, and minimal or absent functional speech. Physiological measures, including electrodermal activity, heart rate variability, pulse, skin temperature, and movement, will be monitored using the EmbracePlus wearable device, while gaze fixation and gaze aversion will be recorded through VR-based eye tracking. Day-care staff will document participants’ behavior and activity context.

Particular emphasis will be placed on the preparation phase for VR use through individualized visual strategies. Participants will receive up to three brief familiarization sessions using visual supports, predictable sequences, and gradual introduction to VR content in calm, low-stimulation conditions. Sessions will last 15–30 minutes and will be conducted only when participants are calm and willing to participate, with predefined criteria for termination. This preparatory phase is considered essential for reducing uncertainty and distress associated with unfamiliar technology, supporting participant safety, and improving the reliability and interpretability of collected data.

The study is expected to identify individual physiological and visual-attentional patterns preceding observable signs of distress. Findings may support earlier recognition of stress, individualized interventions, and the development of assistive warning or biofeedback systems. The study will particularly highlight visual-strategy-based individualized preparation as a prerequisite for safe and meaningful VR-based assessment in people with profound autism.

Power and privilege shape access to resources and decision-making, yet these mechanisms remain hidden in professional practice. Using the Power and Privilege Wheel, this workshop examines how structural characteristics influence the positioning of individuals with developmental disabilities within systems of power, progressively applying this framework to assistive technology (AT) and artificial intelligence (AI). Participants will explore whether these emerging technologies truly expand user agency and meaningful choice, or if they inadvertently reproduce and deepen existing inequalities and dependency.

By the end of the workshop, participants will be able to:

1. Identify how dimensions of power and privilege influence access to resources and decision-making.

2. Apply the Power and Privilege Wheel to evaluate autonomy and choice for individuals with developmental disabilities.

3. Critically examine how assistive technology and AI impact agency, participation, and independence.

4. Recognize technological barriers, such as affordability, digital literacy, institutional structures, and algorithmic bias that contribute to inequalities.

5. Reflect on professional roles in supporting equitable access and meaningful choice regarding AT and AI.

This 45-minute interactive workshop combines brief theoretical input with participant-centered activities. Attendees will first explore the Power and Privilege Wheel through individual reflection and small-group analysis of how structural characteristics impact individuals with developmental disabilities. Next, participants will analyze case scenarios involving AT and emerging AI tools to evaluate who controls resources, makes decisions, and benefits. Comparative exercises will highlight the tension between AI’s capacity to reduce barriers and its risk of reinforcing inequalities through algorithmic bias or unequal access. The session concludes with a structured reflection linking these dynamics to daily professional practice.