We mapped UK-affiliated literature on simulation and AI-dialogue training for helping professions (OpenAlex harvest, July 2026). The headline for social work and safeguarding: the field is small, VR-led, and has almost no UK-led AI conversation simulation.
The numbers
Across 4,646 UK-affiliated records and 46 core works meeting domain × method criteria:
- 22 VR / avatar simulation papers
- 12 AI / LLM simulated-client works (growing from 2021)
- Only 6 specifically about social work or child protection — none newer than 2023
- Only 1 of those six has any AI/chatbot component — and the UK role in that project is minor (Norway-led)
No UK incumbent
No UK group has published an LLM- or chatbot-driven safeguarding-conversation simulator for social work training. Prior art is VR, drama and standardised-client methods — Nottingham Trent, Birmingham, Ulster and UCL/Goldsmiths child-protection recognition work form a lineage to cite, not a competitor set to displace.
That matters for evaluation partnerships: co-publishing rigorous evidence on conversational AI for safeguarding practice fills an empty slot — it is not catching up to a crowded field.
Implications for practice tools
Practitioners still need safe rehearsal for MACPT-style conversations, supervision dialogues and difficult home visits. Text modules and VR suites each solve part of the problem; conversational AI adds scalable, reset-without-shame practice — if grounded in UK statutory framing and evaluated with academic partners.
KallosSim’s social care simulator at app.kallossim.com sits in that gap. HR and employment training lives on this marketing site; safeguarding practice on the product apex.
Method note
Filtering is keyword-based over titles and abstracts with UK institution filter from OpenAlex. Counts for 2025–2026 are conservative due to indexing lag. Full map available to research partners on request.
Social care conversation practice
Voice-first safeguarding scenarios — separate from the HR & Employment simulator.
Open social care simulator →