A values-driven approach to AI in public services
Updated: 5 days ago
Artificial Intelligence has the potential to change public services. An instinctive reaction can be to jump immediately to the possibilities of technology, before we consider the ‘why?’ and ask 'is it safe?'. Andrew Laird argues for starting with a values-driven approach to looking at AI, based on recent work with Newcastle City Council.
Up until about six or seven months ago, you could have been excused for thinking artificial intelligence (AI) was something you could opt out of. You could choose not to engage with this slightly nebulous concept. This is no longer the case. The launch of Chat GPT and other similar tools has suddenly made AI widely available, useful and inevitable. We are all now engaging with AI whether we choose to or not.
This is not to say public services have been unaware of AI. Many council citizen-facing services utilise chatbots to provide 24-hour information to residents e.g. on when bins will be collected. Some have looked into automation to try and tackle the drudgery of admin and paperwork. Fewer still have delved into the world of AI powered analytics to provide increased insight on their residents and services.
Strictly speaking, not all of this is AI. The government’s National AI Strategy uses the following definition: “Machines that perform tasks normally requiring human intelligence, especially when the machines learn from data how to do those tasks.” The key element is the ability to learn by itself – a lot of what is currently referred to as “AI” is nothing more than a decision tree with outputs entirely controlled by the human designer. Automation – yes – AI – not really.
However, with the advent of ChatGPT and other large language models, which actually do have a learning element which is not in our control, public service leaders need to seriously step up their consideration of why and how we use it.
It is not hard to understand why many individuals and organisations immediately jump to thinking about the possibilities of the technology and the products they can buy before considering the “why?” and if it is safe to do so. For a lot of sectors and services, this may be ok and relatively safe. For public services, any adoption of AI has to be deeply rooted in values and ethical considerations.
The Mutual Ventures team are fortunate to be doing some of this initial AI values and principles work with Newcastle City Council. We’ve been working with the Newcastle team to consider the following questions:
How might AI transform the strategic environment within which a council/community operates?
What are the key ethical considerations?
What AI is already being used within the council?
Could AI enhance the effectiveness of council staff by better analysing data patterns and trends to assist with prioritisation and early intervention?
Could some processes be completely automated to create a 24-hour service with limitless capacity?
What are the key practical considerations such as the availability, maturity and appropriate use of data?
What can be done immediately to prepare for potential disruption?
We’ve tackled these questions through the development of a framework which considers both the main use cases for AI in a council or public service organisation and also the key enabling factors. We also carried out a staff survey to establish a baseline for AI use and also general attitudes, hopes and fears. As you would expect the results were varied.
The main use cases for AI can be summarised as: AI powered citizen services; AI powered back-office automation; and AI powered data analytics.
The low hanging fruit are tools like citizen-facing chat bots etc which largely just draw on fixed data sets with no real learning element - and simple back-office automation like transcribing tools. The more challenging element is to establish reliable and non-biased data sets which are large enough to allow genuine pattern recognition. As an example this could be used to spot when a person or family is approaching crisis. The AI tool could then nudge a human member of staff to check in with that family. For me, that would be the big win. Not replacing human judgement or decision making but helping us avoid people falling unnoticed through the cracks until we finally notice them in severe crisis, which is an awful outcome for everyone.
Any organisation wanting to develop a serious AI Strategy and not just jump at the most obvious shiny AI tool also needs to consider the key enabling factors. These are set out below:
We discussed these factors, along with the use cases at the Co-operative Councils Innovation Network conference last week in Sunderland. I led a workshop along with David Hepworth from Newcastle where we described the work we had been doing and some of the challenges and opportunities we had encountered.
The response from the other councils in attendance was encouraging and it seems there is a broad realisation and acceptance that having a waste management chat bot isn’t going to be enough to “tick the AI box” and that a thoughtful values driven approach is what’s needed.
If you would like to chat about your own AI journey then please drop me a line at email@example.com.
Read a case study of Mutual Ventures work with Newcastle City Council here.