The most important kind of AI in local government

On this very recent Yom Kippur, the Jewish Day of Atonement, a day which marks the culmination of reflection and a recommitment to doing what is right, I heard a sermon which featured Artificial Intelligence. The rabbi, who is not a native English speaker, confessed he had been using AI to polish his sermons. He announced that he would write and speak from the heart and stop taking ‘the lazy way out’. While there was much I could take from his message in terms of doing my share of the hard and necessary work, I didn’t agree with him about using AI.

Perhaps as a conscientious rabbi and all-around good guy, he had fewer shortcomings to choose from than I do. Using artifcial intelligence to make my life a little easier doesn’t even hit my top ten. I will certainly be using AI as a writing tool and as an assist for administrative tasks in the weeks and months to come. In fact, I’ll be diving even deeper into AI and looking at how we at LGIU and in local government in general can take advantage of AI tools to deliver better outcomes for local people, ethically and responsibly.

AI and local government: 

More recently, we at LGIU hosted a Global Local event on AI and local governance. It featured Councillor Shanika Mahendran a cabinet member whose brief includes innovation from Milton Keynes, Michael Lawrence Evans, Director of Emerging Technology at the City of Boston in Massachusetts, Balazs Csuvar from DG Cities an innovation consultancy owned by the London Borough of Greenwich and our own chief executive Jonathan Carr-West. The event was chaired by Luke Antoniou Senior Editor at Smart Cities World.

Each of these organisations is making groundbreaking innovations, but they’re also, naturally, wrestling with issues that accompany any radical technological shift. How can we use these tools most effectively? How do we use this tool to make things better without seeming like it’s a solution in search of a problem? How can we make sure that we’re being ethical in how we use AI? The City of Boston has developed interim guidelines on how employees can use generative AI (link to a PDF).

On the very first page:

Generative AI is a tool. We are responsible for the outcomes of our tools. For example, if autocorrect unintentionally changes a word – changing the meaning of something we wrote, we are still responsible for the text. Technology enables our work, it does not excuse our judgment nor our accountability.

The most important kind of AI

I completely agree. I think the most important kind of AI is the Accountable Individual. If I write an article and I use AI to help draft it, I’m still responsible for what goes out in my name. Using AI for drafting and outlining is one of the easier cases, it doesn’t really matter if I understand what’s under the hood of the Large Language Model used to generate text. I can see the product and it’s my responsibility to make sure that the final outcome is something that is helpful, true and informative and not just legible and hopefully a little entertaining.

What becomes harder are the cases where AI is being used to analyse data that inform decisions. For example, using AI to analyse open responses to consultation gives us the freedom to use richer methods of engaging with citizens far more quickly and cheaply than with traditional analytical means. However, this advantage comes with the obligation to use the information to inform our decision-making and to be transparent about how we do so.

Michael from the City of Boston described their work with Google’s Greenlight project. This programme uses AI to analyse traffic data to make recommendations on how traffic lights could be re-signalled to help traffic flow more efficiently. He described how some of the traffic engineering staff were a little nervous about its use, perhaps assigning some of the nervousness to worry about being ‘replaceable’. Certainly, that’s understandable. But what may also underlie unease is that black box computations cannot be held accountable for snarling traffic jams, but politicians and local government staff can be. And while early, but anecdotal feedback suggested improved flow led to less chaotic intersections which improved the pedestrian experience, it’s also true the data set and any recommendations derived from that data didn’t include information on walkers or cyclists. The City of Boston’s transportation policy prioritises those non-motorised infrastructure users, while the automated recommendations could not.

This isn’t an argument for not trying new and promising approaches. Rather it’s an argument for understanding that staff, politicians and citizens may have a variety of reasons for feeling uncomfortable and that as we use new technologies we need to clear-eyed about the benefits, transparent about how we’re using tools and help individuals understand the strengths of weaknesses of any approach so when they are the AI – the accountable individual – they can work with the strengths of artificial intelligence.

Get ready to be the accountable individual

At LGIU, we’d love to hear more about how you’re using AI to reduce administrative burdens and improve outcomes for citizens or about your questions and concerns. We also have more information and resources on the use of artificial intelligence in local government including a recording of the event.

Make sure you’re signed up to our updates either as an LGIU member for full access, including free attendance at future Global Local events or our free updates available to everyone.