By Sascha Giese, Tech Evangelist, SolarWinds
The Government Digital Service (GDS) – part of the Cabinet Office responsible for digital transformation in the UK – is exploring how generative artificial intelligence (AI) could improve the user experience of the government’s website.
Public sector IT experts are using generative large language models (LLMs) – the same technology behind chatbots such as ChatGPT- to come up with a more intuitive way for people to interact with government departments and agencies.
Instead of searching for information – or scrolling through pages of documents – GOV.UK Chat, as it’s known, allows people to simply tap in a question and wait for an answer.
As you’d expect with an experiment on this scale, the results have been mixed.
Early analysis has found that while most people liked using GOV.UK Chat, there were a few teething problems. One of the issues highlighted centred on the accuracy of some of the replies.
As a blog published in January explains, some of the answers “did not reach the highest level of accuracy demanded for a site like GOV.UK, where factual accuracy is crucial.”
It went on: “We also observed a few cases of hallucination – where the system generated responses containing incorrect information presented as fact – mostly in response to users’ ambiguous or inappropriate queries.”
Our AI journey has only really just begun
For me, the findings – supported by another excellent article from the multidisciplinary team running the experiment – are fascinating and provide insights into a project like this.
As they explain: “Recently, there’s been a lot of discussion about artificial intelligence (AI). So, it’s clear the spotlight is on the opportunities and risks AI presents to governments – and what can be done collectively to use AI for good.”
This is good to hear. The rapid pace of development around AI shows no sign of easing. And I get the sense that many people are running just to keep up.
Take regulation, for example. Yes, AI is exciting, and developments are happening at a fast pace. In that regard, I’m pro-innovation – but not at all costs.
That’s why I’m also a huge advocate for regulating AI. Not at some indeterminable point in the future, but now. Why? Because from where I’m standing, I believe we’ve reached a level where if we don’t, things could get dangerous.
At some point, humanity needs to get a grip on AI in all its iterations to ensure the positive effects it could have are realised.
Regulation and greater knowledge are essential to make AI work
After all, ‘AI for good’ – as highlighted by the GDS team – could have plenty of positive outcomes. It could ease the burden on IT teams, for example, by automating routine tasks like password resets and network troubleshooting. Not only would this be timesaving, but it would also liberate employees to concentrate on high-value tasks.
AI used in this way is like having an extra set of hands, but ones that never tire or falter in their accuracy.
But, we must also accept that not all types of AI technology are equal.
When you train AI, for example, it’s like teaching a child – you’re giving it your view of the world. This is why we need people from different backgrounds to train and prepare AI to get a better understanding of our diversity across the world.
I’m sure those working on GOV.UK Chat are finding out about these issues – and many more besides – for themselves. It’s one of the reasons why their work on GOV.UK Chat is so important.
All the lessons learned will help expand our knowledge of how best to implement this technology. This will be particularly important if it’s to address some of the fearmongering around jobs and other issues.
After all, if we’re to make any headway, the general population needs to understand the benefits of AI, understand the privacy concerns, and understand what it means for their daily lives.
Recent Comments