AI has quietly moved from a future concept to a present-day reality in adult social care. Most conversations I have with providers now include it somewhere. There is interest, sometimes excitement, and often a healthy degree of scepticism. Managers are already carrying significant responsibility. The idea of introducing AI prompts a very practical question – will this genuinely help or will it simply create more work?
Recently, colleagues at the Digital Care Hub joined a roundtable hosted by Access Group and Casson Consulting. While this article is not a report of that discussion, many of the themes that surfaced echo what we are hearing more widely across the sector.
When more data does not mean more clarity
Care services are not short of information. Managers oversee care planning systems, medication records, incident logs, staffing rotas, sickness data and feedback from families and people drawing on care. The difficulty is rarely collecting data. It is interpreting it in a way that is timely and proportionate.
One provider told us they did not want another dashboard. That comment captures something important. More charts and more notifications do not automatically lead to better oversight. In fact, they can make it harder to see what is really going on.
Perhaps the more useful starting point is to ask what a manager actually needs to know each day. Are there any serious incidents or safeguarding issues that require attention? Is the rota safely covered? Are medications being administered as planned? Is a member of staff working consistently high levels of overtime? Has someone’s behaviour changed in a way that suggests distress?
If AI can draw together information that already exists and highlight those priorities clearly, it begins to look like support rather than surveillance. The data is not new. The clarity might be.
Spotting patterns and preventing crises
We are hearing practical examples of how AI is being used to identify patterns that would otherwise be difficult to spot. In one area, historic shift data was analysed to understand which care workers were most likely to accept particular slots. The system was reportedly around 95% accurate in predicting whether a shift would be picked up. That does not remove the human element of rota management, but it can reduce last-minute pressure and make conversations about preferences more grounded in evidence.
Another provider described using data more consistently to monitor changes in behaviour in a learning disability service. By analysing patterns over time, they could see that certain individuals were more likely to struggle at particular times of day. Instead of reacting after incidents occurred, they adjusted support proactively, increasing presence where needed and easing off when things were more settled. The aim was prevention, not control.
A particularly striking example involved a young autistic man who had been receiving 210 hours of intensive support each week due to behaviour perceived as challenging. Through careful monitoring and thoughtful analysis, the service began to question whether constant supervision was helping or, in fact, contributing to social burnout.
The data suggested he needed more time alone rather than more oversight. Using technology alongside professional judgement, and with consent, the team gradually reduced support to 10 hours per week. Distress levels dropped significantly and his quality of life improved.
As Sinead McHugh-Hicks, Managing Director of Dimensions, reflected, ‘Using AI and technology has supported us to take informed, positive risks which has in turn enabled us to reduce the restrictive impact of excess supervision and social burnout often experienced by autistic people, leading to less behaviours of distress, reduced hours of direct support and increased quality of life for people.’
In this situation, better support meant less intervention. Data provided reassurance that a positive risk could be taken safely, but it was the staff’s understanding of the individual that shaped the final decision.
Getting the foundations right
Examples like these also highlight the importance of how data is recorded. Consistent keywords and coding across systems allow AI to pull together meaningful insights from narrative reports. Recording preventative interventions, not just outcomes, gives a more balanced picture of what is happening in someone’s life.
Closed AI systems can reduce certain risks, particularly around data security, but they still depend on high-quality, accurate information. Bias and error remain real concerns if systems are poorly designed or fed incomplete data. Technology cannot compensate for weak recording practice. This is where governance and clarity of purpose matter. AI should be introduced to solve defined problems, not simply because it is available.
Nothing about us without us
Trust underpins all of this. People drawing on care, family carers and staff need to understand how their data is used, what is personally identifiable and what is not and how AI-generated insights feed into discussions and decisions. Consent must be clear and ongoing.
The principle promoted by Think Local Act Personal, ‘Nothing about us without us,’ feels particularly relevant in this context. In the London Borough of Enfield, residents were involved in developing a chatbot designed to improve communication. Their input shaped both the design and the language used. When people are involved from the outset, technology feels less like something imposed and more like something shared. Co-production does not remove
all concerns, but it does build confidence and surfaces important questions early.
A wider ethical conversation
There is also broader work underway to explore these issues in a structured way. A joint project between the Institute for Ethics in AI at the University of Oxford, the Digital Care Hub and Casson Consulting is examining the ethical use of AI in adult social care. Care providers, technology suppliers, people drawing on care, family carers, care workers, commissioners and regulators are all involved, reflecting the reality that these decisions affect the whole system.
Regulators are part of the picture too. The Care Quality Commission has indicated that it does not wish to curtail innovation, including innovative use of AI. At the same time, providers must be able to demonstrate that any technology used is safe, proportionate and aligned with good care.
Some organisations are establishing ethics committees or governance groups to oversee technology and AI across the whole service. This goes beyond workforce analytics. AI may sit within HR systems, care planning tools or monitoring platforms, and oversight needs to reflect that breadth. The aim is to ensure that technology prompts reflection and discussion rather than quietly automating decisions.
Keeping care human
Social care remains, at its heart, a human endeavour built on relationships, empathy and understanding. AI cannot replicate those qualities and it should not attempt to. What it can do, if introduced thoughtfully, is help managers see more clearly, feel less overwhelmed and take informed decisions with greater confidence.
The challenge is not whether AI belongs in social care. It is whether we shape it in a way that supports managers and strengthens people’s lives, rather than adding to the complexity they already navigate every day.
For further information, including guidance on the ethical use of AI, visit the Digital Care Hub website. Comment on this feature below to join the conversation to share your thoughts.
Katie Thorn is Director of Innovation at Digital Care Hub.
Email: [email protected] Linkedin: @Digital-Care-Hub