Robotics revolution: why chatbots and AI could shake up local government
Enfield Council’s lead technology adviser Rocco Labellarte says the authority’s chatbot is attracting international attention and is about to transform services and release resources. Gill Hitchcock reports
IPsoft's flagship cognitive agent technology is called 'Amelia' Credit: IPsoft
Rocco Labellarte has been on a journey of discovery.
Over the past year, Enfield Council’s lead technology adviser has travelled from researching the potential of chatbots and artificial intelligence (AI) in local government, to preparing to launch a chatbot, or “cognitive agent”, to interact with citizens.
After four years heading up IT at the Royal Borough of Windsor and Maidenhead, Labellarte joined Enfield in 2016 – just one week it had agreed a research and development partnership with US AI company, IPsoft, to build this new cognitive agent.
A number of suppliers had approached the north London borough of Enfield to promote AI as part of its digital transformation programme. When the council looked into the AI market, however, it became clear this was an immature technology.
- London borough Enfield enlists AI employee
- iCouncil: Do robots have what it takes for local government?
- Enfield seeks to extend Serco ICT service contract, rather than retendering
But, Labellarte says, it’s never too early for research and development into technologies that have the potential to benefit both staff and citizens.
“With that premise, we started the ball rolling,” he adds. “Fast forward a year, and we have been working very, very closely with IPSoft, but also with Microsoft, as the technologies mature.
“Now we are at a point where we are probably weeks away from deploying the first real, effective cognitive agent – certainly into beta testing with a wider audience, but potentially into the public domain as well.”
Extensive press coverage at the time the IPsoft contract was announced jumped the gun by reporting that Enfield was launching a robot, or avatar. The R&D angle was largely ignored.
“I am going to tell you a little anecdote,” says Labellarte. “This time last year, when they announced the signing of the contract, it went viral. We had live chat with real agents on the website and several people complained that they didn’t like the fact that they were talking to a robot, when they were actually talking to real people. It’s interesting how perceptions and reality start to blur.”
While the first avatars, which can converse in multiple languages and respond to a wide range of queries are just starting to be produced, Enfield has been working on a text-based chatbot for planning services.
“It has the potential for spoken commands, so you can speak into it as well, but it is quite a long way from having an avatar or virtual assistant that walks and talks,” says Labellarte.
Why did Enfield start with its planning department? Labellarte says all councils, whether a district, borough or metropolitan authority, have three type of services. There are high-volume, low-complexity transactions, such as paying council tax or reporting a pothole. For these, Labellarte says, AI adds a layer of complexity and electronic forms are a better way forward.
At the other end of the spectrum, councils have low-volume, high-complexity services. Adult social care, looking after children, and vulnerable adults are among them. Right now, AI, robots and chatbots are not ready to have an interactive conversation with people who are vulnerable.
“They need the human touch,” says Labellarte.
What is left are the mid-volume, mid-complexity transactions, such as planning services. The surveyors and others who staff these services can be expensive and, after a pilot with 45 planning permission and building control processes, this is where the cognitive agent will be launched.
‘Is that a real person?’
Labellate describes how the service could work: “It’s Saturday evening, it’s 10 o’clock and I am having a conversation with my partner about how it would be really good if we could extend our property.
“What do we need to do? To find out, we go to Enfield’s website, click on ‘planning permission’ and they say ‘would you like to talk to our virtual assistant?’ I click ‘okay’, then I immediately have an interaction which feels like a live chat. And the interaction is such that you ask yourself, ‘is that real person?’.”
Of course, the project could put jobs at risk, but then councils are already automating services in an effort to save money by cutting staff. Since 2010, funding cuts and increased demands on services because of a growing population meant that Enfield had to make £118m million savings, with £31.6m of these made in 2015/16, according to its accounting statement for that year.
It’s the most exciting field in technology within the last 15 years
Labellarte argues that cognitive agents will free up resources so Enfield can provide services that would otherwise be unaffordable. And he says that AI, natural language semantic processing, machine learning, visual recognition, and other emerging technologies require a new set of skills. The council will look to develop those skills internally, creating new opportunities for staff.
While he is reluctant to reveal Enfield’s financial outlay on the project, he says it will cost about £3,000 per process which is “minute”.
Enfield has been spreading the message too and, over the past year, has been talking about the project to about 30 other councils.
“We have been approached not just by councils here, but by the City of Hamburg, and a major international bank who have come and seen what we are doing,” he adds.
Asked about the biggest challenge, Labellarte responds: “It is natural language processing, the ability of a system to take a phrase and understand it. Chatbots, you tell them what your question is, you get possible answers and then you say ‘yes’ or ‘no’, and the chatbot has to work out whether you have said yes or no, red, blue, or green.
“With our cognitive agent, people can input text, and where it is enabled they can speak into it, and there are buttons they can use as well.”
Labellarte says it’s still early days, but as this technology matures it will create a revolution across all sectors. And that revolution, he thinks, is only about six to 12 months away.
“It’s the most exciting field in technology within the last 15 years,” says Labellarte. ”Forget cloud, forget software-as-a-service, forget all those things, this is undoubtedly the most exciting new technology.”
The health service needs to make sure its people are geared up for the data-rich future, according to Afshin Attari of Exponential-e
Trials also ongoing in local government and private sector
Report from National Audit Office claims public sector must do more to combat problem
Outage in early hours of 2017 saw call-handling and dispatch system fall over