Skip to main content

How AI is used in the public sector

Blog-Image-Blue-Ai-public-sector

This blog has been expertly reviewed by Andrea Rosales, Lead Data Scientist at Colibri Digital. 

Artificial intelligence (AI) is now mainstream, with tools like ChatGPT impacting almost every area. The private sector has seen a significant shift, rolling out new AI systems that improve employee efficiency, create new products and increase profits. 

But one area poised for further AI potential is the public sector. A recent report estimates that governments could use AI to free up 30% of civil servant time by automating mundane tasks. Likewise, AI tools can create new systems that benefit both providers and citizens. 

With its robust regulations and duties of care, public services must take the utmost care when implementing new technologies. While the potential benefits of AI are massive, rolling out new systems is no easy task. The sector is hugely complex. Strict quality checks must be in place, and, with cyber security a constant concern, any new deployment must safeguard data to the strongest possible degree. 

However, with so much to gain from AI, future implementations are inevitable. The UK Government recently released its National AI Strategy, aiming to use AI to fuel productivity, safety and growth for citizens. In this blog, we’ll see the types of systems the public sector could use, how it already uses them, and how it can prepare for the future adoption of AI. 

Andrea Rosales, Lead Data Scientist at Colibri Digital, said: “As a data scientist, I see AI as a catalyst for innovation in the public sector. Driving smarter governance and more effective public services through cutting-edge data science and innovative technologies will only enhance the well-being and prosperity of citizens.” 

What are the potential benefits of AI for the public sector? 

Each AI model brings potential use cases and benefits for the public sector. 

  • Cost and time savings: One of the top potential benefits of AI is automating mundane, error-prone tasks. For the public sector, this may mean automating documentation-based work to free up civil servant time — all while increasing accuracy, too. 
  • Accurate decision-making: AI can also help public services make more informed decisions. Any system involving lots of data and human time is a good candidate. 
  • Deeper insights: Governments can use AI models to spot patterns within their data. This may help fuel improvements or minimise inefficiency.  
  • Improvements to public services: The ongoing aim is for AI implementation to reach the end users, improving services and daily life for citizens. 
  • Easier embracing of emerging technologies: Digital transformation requires a culture that’s ready for it. By testing and implementing new technology, the public sector is preparing its workforce and infrastructure for future developments. 

UK public services have already realised some of these benefits, while others are ready for further innovation. 

Public services AI use cases 

1. Government services 

In early 2024, the UK government trialled a generative AI chatbot on its gov.uk website, answering common visitor queries. The initial feedback was positive. Almost 70% of users found the information they needed with the chatbot’s help. Two-thirds of users were satisfied with the experience. But, as with all AI implementations, there were issues with some mistaken outputs, including hallucinations and incorrect information. 

With more trials planned, the gov.uk chatbot could be rolled out permanently. That would free up civil servants' time while providing users with faster answers to their questions. 

2. Healthcare 

The healthcare sector has enormous potential to use AI in many ways, such as: 

  • Accurate care: The NHS has already used AI to improve care, such as an AI tool that uses image recognition to detect heart disease up to 30 times faster than a human could. 
  • Efficient service: Using NLP and machine learning, healthcare could rid itself of many laborious admin tasks. 
  • Proactive service: AI models and Internet of Things (IoT) devices can provide more specialist, remote care to those who need it. 

3. Public safety 

A recent Deloitte study highlighted that smart, AI-driven technology could help reduce crime by a third. The same study predicted that automation could also improve emergency response times by 20%. 

There are many ways it could do this.  

  • Image recognition could help services spot real-time suspicious activity quicker, such as those already used in Wales and at large events. 
  • Machine learning could predict crime hotspots for future prevention. 
  • Models could actively monitor for cyber threats. 

4. Education 

Generative AI has massive potential in the education sector. The UK government has already invested millions of pounds in researching AI-powered teaching tools, while the Department for Education released a report on using AI within the field. Almost half of the teachers surveyed have already used AI in their roles. 

Future developments could help with tasks from lesson planning and marking to producing educational materials. Interestingly, AI may also be needed to spot the use of AI within students’ work. 

Andrea Rosales commented: “We’re seeing more and more educators and education providers asking for support with implementing AI solutions, with a variety of reasons cited. For some, it’s a response to market changes, while others want to promote cost savings and data sovereignty. Personally, I think we’re going to see AI grow tremendously in this area of the public sector in the coming years.”  

5. Transport 

One of the most compelling uses of AI is to reduce errors. Up to 95% of vehicle collisions are caused by human error. Everything from autonomous vehicles to intelligent motorways may help improve public safety when using public and private transport. 

We’ve already seen AI models helping to improve motor safety. The Department for Transport used machine learning to look at patterns in MOT data — highlighting providers that gave inadequate or unsafe vehicle inspections. Likewise, a new AI model is being rolled out to spot hazardous driving, including those not wearing seatbelts or using a mobile phone while driving. 

6. Climate 

Finally, AI can help governments with their big-picture ambitions, including meeting climate challenges. The UK government has granted millions of pounds to AI research to reach green targets, while other models can help spot carbon emission hotspots. Finally, forecasting models may help pinpoint potentially catastrophic weather earlier, giving people and businesses more time to prepare. 

Types of AI systems  

While its roots go back decades, modern AI has accelerated in the last few years. Faster processing speeds combined with cloud computing infrastructure means organisations have more access to AI-ready tools than ever. 

AI comes in various forms. Some systems produce new information, some make estimates, and others highlight problem areas. But all of them begin by learning from past data. To make their decisions, AI models take data, understand patterns in that data, and then produce results based on their findings. Where they differ is in their outputs. 

  1. Generative AI: OpenAI’s ChatGPT has shown the world the potential of generative AI, using its learnings to produce new information based on the user’s input. Generative AI can create new images, text, and even music. 
  2. Machine learning: Machine learning is at the heart of AI. Computers learn patterns in data and then act on their learnings to manage, improve or automate future tasks. 
  3. Natural language processing (NLP): AI algorithms can use NLP to find patterns in text. While generative AI uses that learning to create new text, NLP models might use it to translate, transcribe or understand documents. 
  4. Computer vision and image processing: Learning from thousands or millions of example images, AI models can spot the tiniest patterns in future images. We see this used often in healthcare and self-driving cars, for example. 
  5. Deep learning: Deep learning algorithms are built on neural networks, most commonly deep neural networks (DNNs). These mimic the human brain to make incredibly complicated decisions — often well beyond human capabilities. 

Considerations when using AI solutions 

With such powerful use cases, it would be tempting to think that AI is the answer to every problem. But it only works if the correct factors are in place. These include: 

Quality of datasets 

The entire value of AI is built on having high-quality, relevant data to learn from. If that is not available, the results will be worthless at best, and harmful at worst. For an AI project to succeed, data quality is non-negotiable. 

Andrea Rosales, Lead Data Scientist at Colibri Digital, said: “Data quality is of the utmost importance, poor or biased data can lead to inaccurate or biased AI outcomes, rendering it practically worthless. It’s also important to consider bias and fairness to ensure the model does not disproportionately affect any group negatively.” 

Budget 

It’s also important to remember that AI can be costly in the first instance. Models take massive amounts of computing power to train, requiring large on-premises or an experienced cloud services partner. However, AI is also thought to reduce costs in the long-term so it’s all about considering your budget and costs both now and in the future. 

Ethics 

Of course, any AI model must meet ethical requirements. This goes a step further for the public sector, where protecting the interests of citizens is of the highest concern. Any AI model must meet the right policies, such as GDPR and data protection acts, while minimising any potential biases or errors in its output. 

Explainability 

A primary concern with using advanced AI is that humans cannot comprehend the calculations used within models. Simply put, in many cases, we don’t know how an AI model came to its decisions. If using a model for anything that could potentially cause harm, it’s essential that there is a measure of explainability about the decision-making process. 

Security 

The data used to train AI — and the subsequent results — must be protected from breaches. This means considering factors such as access protection, encryption, meeting regulatory needs and the secure connection with legacy systems.  

How Nasstar can help 

AI has unlimited potential for enhancing government services. Across various sectors, AI models can automate error-prone tasks, find new insights and open up opportunities for innovation. But they will only work if the proper considerations are in place, including security, infrastructure, expertise, fairness, and high-quality data. 

The Colibri Digital team, part of the Nasstar Group, can help your organisation build an AI-ready infrastructure. With cutting-edge technology and an expert team, we can help you design, develop and secure an environment ready for innovation. 

Speak to a specialist to learn more. 

Frequently asked questions (FAQs) 

Can AI improve public sector productivity? 

AI can help improve public sector productivity across many areas. One example is automating mundane, error-prone administrative tasks, such as those currently on paper files. 

How could AI be used in public services? 

AI could be used in public services in several different ways. We may see generative AI produce lesson plans within education. Likewise, image recognition could help healthcare providers spot disease quicker, or help police monitor significant events for dangerous activity.  

What are some AI in public services challenges and opportunities? 

AI has the massive potential to streamline public services while creating new opportunities. However — as with all AI systems — it must be built securely and fairly, based on high-quality data, with the end user in mind at all times.