Artificial Intelligence (AI) – Next wave of digital for the public sector?
What is Artificial Intelligence (AI)?
Computer scientists since the late 1950s have being developing software systems in order to explore the validity of the following proposition:
“Intelligence and learning qualities can be replicated or simulated in a machine (computer) so that we can build intelligent entities”.
To clarify, in Artificial Intelligence and computer science, “intelligence” is represented by qualities such as:
- Performing the correct human-like and/or rational action or response when given some information (e.g. undertaking an assessment based on findings or information)
- Ability to recognise a pattern (e.g. identify a friend when lost in a crowd)
- Operate as an autonomous agent (e.g. navigate a warehouse to locate and recover an item).
The aim of AI is to uncover the principles of intelligent behaviour, including learning, and then attempt to replicate these in a computer or automated system. However, the aim of AI is not necessarily to fully replicate the biological mechanisms that underpin human intelligence. The nearest analogy to the mission of AI is the comparative flying qualities of birds and airplanes. We can fly using airplanes, utilising the same physical principles as birds use but we employ a totally different underlying mechanism.
So, the practical results of AI research are that if we understand the underlying principles of intelligence and learning, we will be able to find significant utility from deploying AI solutions. The good news is that many applications have already been found for AI in the commercial world.
Historically, many academic disciplines have contributed research to uncover the principles of intelligence and these have helped progress the mission of AI:
- Control Theory
Why is Artificial Intelligence important now?
Even though the research to replicate intelligence in computers started very soon after modern digital computers were invented, the processing power, data storage capacities and communications bandwidth have only now reached a capability to make AI solutions operate reliably and quickly. Apple’s Siri, Google’s Voice Search and car GPS navigation systems are all examples of the application of AI principles and solutions to create helpful digital systems.
With the near universal availability of powerful modern Smartphones, we are now in a position to exploit the potential of AI to drive further transformation and improvements in public digital services.
The opportunity is to incorporate in to computer software routine processing rules that would be followed by public servants applying policies and procedures when assessing the eligibility of services for the public. Citizens could undertake these assessments themselves when it is convenient and thus save unnecessary appointments and face-to-face meetings.
The other opportunity is to use the vast amount of data about patients and social care clients to identify common patterns in order to offer preventative interventions and better personalise services to patient/clients.
Where could we use AI for public digital services?
As a result of the Government’s Digital by Default strategy, many more Government services for the public and businesses now include a digital service option. On the whole, the convenience of digital services and the associated transparency, if implemented well, represents a compelling benefit for users. At the moment these digital services have had limited functionality; the main focus being transactional services or online forms submissions (e.g. online tax submission, updating driving licence details, application for student loans, etc). It is now possible, with the addition of AI capabilities, to reduce the workload for Government departments and also help the citizens undertake certain tasks at their convenience and this avoid unnecessary travel and appointments. Here are some examples that would benefit patients/citizens and the professionals who deliver public services using two common AI techniques of rules based processing and pattern recognition:
- Self-diagnosis for common ailments or problems. The diagnostic data once captured, could reduce face-to-face (or video) time and improve the clinical decision making with a health professional
- Self-assessment for benefits / support services / grants / subsidies. Again, once the data and assessment have been captured digitally, future follow-up services can be more efficient and effective.
AI pattern recognition techniques (also referred to as learning algorithms) can also help with the effective management of services and planning for future demand:
- The use of health and non-health data, to identify early indicators for chronic diseases such as diabetes and then use this analysis pro-actively to identify patients who could benefit from early intervention
- Assisting decision making and planning for professionals, especially in social care settings, by automated comparisons of the circumstances and needs of a client to a larger set to identify the successful interventions and actions from historic cases (e.g. covering health and social care).
AI represents an opportunity to reduce costs for delivery of services, improve the efficiency and effectiveness of service delivery and also has the potential to improve the quality and consistency of services delivered.
What should I do to exploit AI for my organisation?
The steps to follow for AI development are:
- Long list of opportunities – identify areas where there are difficulties in addressing demand for services because of capacity issues.
- Viable list of opportunities – from the long list, identify those services that require expertise, expertise which is well understood, such an assessment task or diagnostic process. Also identify diagnostic assessments where there exists body of data about historic assessments (these would be needed for solution that would use pattern recognition techniques).
- Short list of opportunities – from the viable list, identify the likely benefits of automation or part automation using AI. The benefits would be concerned with increasing capacity for highly skilled professionals so they can spend time on more complex tasks, improving the consistency in service delivery to reduce rework and reduce unnecessary work and making the service available (or part of the service) at a time more convenient for patients/clients/citizens.
- Business case for a pilot – pick the most promising opportunity and make a case for seed funding to develop a prototype the codification of the expertise or application of machine learning algorithms (pattern recognition).
- Develop prototype and review – at the end of the prototype phase, review how to best take forward the opportunity. The typical life-cycle of an AI solution is to use it in-house first, for example, to train less experience staff (e.g. junior doctors). And then to use the feedback from in-house applications to package for release as an external public/citizen facing service.