Perspectives
AI in social care: where it helps, and where people must decide
Pelp Care · · 6 min read
Artificial intelligence is appearing in more and more care software. Used carefully, it can reduce administrative work and help care teams notice things they might otherwise miss. Used carelessly, it can create confidence in information that has not been checked.
This article sets out where we think AI can genuinely help in social care, where caution is needed and the principles we are building Pelp around.
Where AI can help
Care teams create a large amount of information, and there is rarely time to read and compare all of it. AI is well suited to organising that information and making it easier to work with.
- Tidying spoken or typed notes into clear, structured records
- Turning an authorised assessment conversation into a draft for review
- Summarising recent notes before a review or handover
- Highlighting recurring themes or changes from someone’s usual pattern
- Answering questions about authorised records, with links to the sources
Where caution is needed
AI systems can produce text that sounds confident but is wrong or incomplete. They can miss context that a support worker who knows the Individual would notice straight away. They do not understand people in the way that care professionals do.
For that reason, AI should not make clinical or professional decisions, assign risk scores without explanation or present predictions as facts.
Principles we are building Pelp around
We believe AI in care should be visible, reviewable and accountable. These are the principles guiding how we are designing Pelp Intelligence.
- AI generated content is clearly marked as a draft until a professional reviews it
- A person, not the AI, decides what becomes part of the official record
- Insights show the records they are based on, so they can be checked
- AI only uses information the user is authorised to see
- Patterns are presented as prompts for attention, not as diagnoses or predictions
Questions to ask any supplier
If you are considering care software that uses AI, it is worth asking a few direct questions before you rely on it.
- Which features use AI, and what data do they use?
- Is AI output reviewed by a person before it is saved?
- Can you see the sources behind a summary or insight?
- How is personal data protected when it is processed by AI?
- What happens when the AI gets something wrong?
Keeping people at the centre
Care is about people supporting people. The value of AI in social care is measured by whether it gives professionals more time and clearer information to support the Individuals they work with, not by how much it can do on its own.