Exploring AI’s potential for public services
Principles of fairness, accountability and transparency will underpin future developments of artifical intelligence
Frank O'Donnell, head of public sector, Microsoft: "We are seeing an increasing use of AI technology in healthcare." Photograph: iStock
Artificial intelligence (AI) has the power to bring about dramatic improvements in public services, but only if it is implemented in the right way, according to Microsoft’s head of public sector, Frank O’Donnell. “It has to be done in a way that is fair, transparent, reliable and upholds privacy,” he says.
He also points out that AI is not really all that new. “AI has been around for a long time, but it is only in the last couple of years that we have had the computing power, the data and the intelligent algorithms to make it work.
“But it is critically important that we adopt AI because of the positive way it can impact on public services and government. It is also vitally important that we move at the right pace to be able to capture the benefits for citizens.”
He mentions a few examples where AI is already having a profound impact.
“We are seeing an increasing use of AI technology in education. Students are using technology to learn, and that technology can then learn their studying style and their strengths and weaknesses and tailor content delivery to meet their individual needs. It would be very difficult for a teacher to do that with a mass audience of 20 or 30 students, or a lot more in a lecture theatre.”
Another good example is healthcare, where image recognition technology is being used on scans and X-rays. “When AI is applied, millions of images can be scanned and patterns underlying diseases and conditions detected,” O’Donnell notes. “This can be used both as a diagnostic aid and to assist in predicting outcomes. This is already happening in some countries and we need to keep pace with that.”
Dr Austin Tanney, head of AI with Kainos, Microsoft UK Partner of the Year, has already seen this in action. “I come from a pharmaceutical and healthcare background and we used AI to analyse genomic information,” he says.
“We used it to understand patterns. That’s where AI comes into its own – understanding patterns in datasets which are too big for human capability. We used machine learning to improve cancer diagnosis. The better understanding of certain cancers and sub-types has led to improved cancer treatment and much of that is due to machine learning and AI.”
In public services more generally, O’Donnell explains how AI can make them more accessible to citizens. “It can improve the way citizens engage with government online or over the phone.
“If you want to know how to renew your passport or driving licence you will have the ability to use chatbots which understand and use natural language. You will be able to type in a standard sentence like ‘how do I renew my passport?’ The bot will know about you and your details and will be able to tell you what you need to do for your particular circumstances.”
AI in action
Belfast-based Kainos recently took the top honour at the UK IT Awards for a project it implemented at the British Driver and Vehicle Standards Agency (DVSA). “This is a fantastic example of AI in action,” says Tanney.
“In the UK, the MOT vehicle roadworthiness test is carried out by independent garages which are licenced by the DVSA. There was a fear that there could be fraudulent behaviour in the system. We used machine learning to detect patterns which identified garages which were being less than scrupulous. As a result, a number of garages no longer carry out the tests. The public benefits from this in terms of improved road safety.”
The same principle applies to the detection of anomalous activity which helps prevent identity theft, he adds.
There must be clear principles focused on fairness, accountability, inclusivity and transparency when using or creating AI technology
The citizen has to be at the centre of this new technological revolution, O’Donnell insists. “It is very important that the paradigm is one of assisting and amplifying human ingenuity. It must be about machines working hand-in-hand with humans, helping them make more informed decisions.
“Indeed, if you look at the economies where robotics and automation are most advanced and utilised, they have lower unemployment than those which aren’t adopting them.”
He stresses the national importance of this aspect of the technology.
“Government can’t be complacent, it has to help all of us move into an era where AI is more pervasive by reskilling people and help them benefit from it. It also has to be a guardian. We need the right regulatory and ethical environment to ensure it is used for the right ends. It is about using AI in the right way.
“There must be clear principles focused on fairness, accountability, inclusivity and transparency when using or creating AI technology in order to respect the privacy of individuals and to help others build systems that are reliable and safe,” he continues.
“We have a team focused on that here in Microsoft. With an ethical framework in place and a human-centric approach, we can ensure that AI has a positive impact today and for future generations.”
Bringing democracy closer to citizens
AI is being deployed in the Houses of the Oireachtas in ways that aim to bring greater transparency and accountability to the democratic process.
We are opening up the good work being done in the Houses of the Oireachtas. We are using AI to make it more accessible to the public
“We are on a digital transformation journey,” says chief information officer Finn de Bri. “We are working on a number of areas at present. In some cases, we are actually using the technology, while in others we are still in exploratory mode.”
Publishing the written transcripts of debates online is one area where AI is going to be looked at. At present, the objective is to have all debates online within two hours. However, each 10-minute recording can take up to two hours to transcribe.
“We are investigating the use of AI for speech recognition to assist this process,” says chief technology officer Ciarán Doyle. “We are using it for natural language processing. Over time we can train the technology to learn the lexicon of the Irish parliaments and understand the dialects and accents of different speakers. It will deliver a significant saving in time and effort .”
“We are using this assistive technology to make the publishing process more efficient,” says de Bri.
The next stage of the project will see every piece of speech and video digitally tagged. “If you search for words in a speech it will be able to find the text and the video.”
Facial recognition software could be used to tell precisely who is speaking and when.
“We also plan to use the Microsoft AI-powered translation tool. People will be able to translate the debates into the language of their choice. Our ultimate aim is to have all European languages, including Irish, available to people. If someone from Poland is looking at an Irish parliamentary debate on an EU treaty they will be able to have it translated into Polish in real time. Microsoft has a suite of other cognitive tools and our goal is to use them all.”
“We are opening up the good work being done in the Houses of the Oireachtas,” says Doyle. “We are using AI to make it more accessible to the public. This is a big opportunity for the members, for democracy, and for citizens. It’s very important for citizens to see what they are getting for their vote.
“It will enhance the democratic process by creating much greater transparency and offering people real-time access to their parliament and elected representatives. This is very important in the age of fake news, when people need a way of knowing what is true and what is not.”
Read more about Microsoft’s research into AI in Ireland.