hands up people who’ve ever said “i cant wait for my local council to build an AI chatbot”
thinking about AI. by which i mean LLMs. by which i specifically mean public-facing chatbots and that sort of stuff.
do residents actually want this?
wouldn't they rather use their chosen AI - which already has access to and understands the internet, their preferences, their context? instead of our RAG bot that knows nothing about them and only works within our little bubble?
see also: council website maps
there's a pattern here. most people don't rush to council website maps when they are looking for services nearby. they don't care about everything being organised around "The Borough".
they use google maps because its centered around THEM. they don't need to tell it where they are - it already knows. it shows what's open right now. it shows them everything relevant, regardless of which council boundary it sits within. it understands context. it prioritises the user over The Borough.
the rush to "do" AI
there's a big push to "do" AI because everyone else is. the key outcome seems to be "doing something AI" over considering the actual benefit for users or how it integrates into their experience.
is it a waste of time for councils to rush to develop this stuff?
one of the main benefits of natural language chat interfaces is / will be the understanding of context and implied meaning. not just the question being asked, but the wider context of a user's life.
like the google maps example - i open the app and instantly everything is relevant to my location and what's open right now. i don't need to spell that out as part of my query.
by the time we get there, everyone else has moved on
we're often slow to react to new tech because we focus on what others are doing. which, if you're looking at other councils or govt departments, is already behind the actual bleeding edge.
then it takes so long for the knowledge to filter through our layers of bureaucracy that by the time it gets actioned, it's way out of date.
with something like this we should be looking at the core capabilities of the tech itself. and the general direction of travel for it in the context of how the public actually experience and use it.
not just copying what other govt departments are doing, because they're probably working from the same delayed playbook we are.
what if we focused on the plumbing instead?
if i want to do a task or find information, i can use the AI service i pay for / use regularly. it has loads of context about me. plus it can search everything available on the council website anyway.
so maybe we should concentrate on publishing readable data instead? maybe we need to jointly work on a council MCP server, API type functionality?
make our stuff accessible and understandable for the AI tools people are already using, rather than trying to build our own versions that will always be worse.
“doing AI” or being “AI first” doesn’t have to literally mean training models and building chatbots. it can mean being aware of and integrating with the wider technology landscape, keeping up to date with developments, really considering how this tech will affect our residents lives. and how we can influence those effects to be positive.
but, particularly as big tech gets bigger and more advanced, that won’t always be feasible to do by controlling and building everything ourselves.
the privacy trade-off
controversial take maybe - but we do need to accept that there is a trade off between privacy and personalised experiences. not all data usage is evil and wrong.
products that hit that sweet spot will be successful for the majority of regular users, who value simplicity and ease of use.
the question is whether councils building isolated AI chatbots is actually serving that, or whether we're just building another thing nobody asked for because we feel like we should.