Opportunities for AI in social service systems

AI is a transformative technology that is still in its emergent phase. While there are risks to manage, social service organisations can benefit from it, particularly with a focus on inefficient processes.

Risks

We can't talk about AI without some discussion about its risks. At an economy level, there are risks that AI replaces many jobs but doesn't create new ones. This of course could create new pressures on social service systems in increased joblessness and related social problems. There are also problems with accuracy (so called 'hallucinations'), which some researchers think are intrinsic to AI and won't be able to be eliminated entirely. There are also worries about public AI models training on private organisational data, although this is a problem that can be solved by good governance and set-up of AI within organisational systems.

In the social sector things are different than in the commercial economy - the demand and need for social services tends to exceed our resources and ability to deliver services, so there is always a demand gap that needs filling. So improvements in productivity and efficiency in the social sector can always be applied to expanding the number of people we help with a given dollar investment. This can then enable organisations to shift staff time to more impactful direct client work. We know of social organisations that have already reduced administrative roles in HR with the help of AI, in order to allocate more resources to frontline expert staff.

"Doing more with less" is a day job for social sector managers, and data and AI are tools that can be used to help this, as long as the risks are well managed.

Opportunities

There are many parts of the social sector that are inefficient. We know of plenty of examples where staff are required to enter data in one Client Management System only to have to duplicate it manually in government databases and spreadsheets. Managers often have data sources spread out and spend many hours cleaning and collating data for their reporting requirements. Many hours are spend doing administrative or data entry tasks, or collecting data that never gets used. For some roles, this burden could be as high as 30-40% of their time.

Many organisations use combinations of physical written documents, PDFs, Word documents, spreadsheets and CRM/CMS systems to capture information. Making these systems work together is time-consuming and often very inefficient. It wastes valuable frontline staff time in non-productive administrative tasks.

It's these duplicated, wasteful tasks that good data design and AI can improve. It's not just putting an AI agent on a task, it's also about thinking about the process or job that is being done and what data is used and fed to AI to do that job. External experts such as consultants can be helpful in providing an independent view of a work process to simplify or improve it.

Our recommendations -

There's still lots to learn by running careful experiments with AI, but it makes sense to start with the 'low hanging fruit' of improving efficiency in your systems.

• Recognise that staff are probably already experimenting with using AI in their jobs even if there is no organisational policy (BYO AI using open web tools such as Chat GPT and Claude)

• Focus AI efforts on efficiency and productivity enhancements rather than on replacing the creative, personal or value-added part of staff roles

• Conduct a review that considers the business process / service design along with opportunities for using AI in conjunction with other data and automation tools - AI is less useful 'by itself' and more useful integrated to do very specific tasks (AI 'agents')

• Consider areas for AI pilots such as improving data entry, summarisation of information, triage, categorisation of tasks and files based on your program logic, removing duplication of administrative tasks

• Embed the AI within the tools that your staff use rather than using them separately - and ensure the process and data produced by AI is auditable and transparent within the system - i.e. so senior staff can override or edit it.

Get in touch with us if you want to explore some ways to improve efficiencies in social service delivery.