Is the public sector implementing Artificial Intelligence properly?
Artificial Intelligence provides huge opportunities for government to deliver more for less through increased productivity and to improve the lives of citizens.
The government understands its importance – AI and Data was named as one of the four ‘Grand Challenges’ in the Industrial Strategy White Paper.
PwC estimates that AI could contribute $15.7tr to the global economy by 2030. The UK is in the top three countries globally in the development of AI technologies. And AI could increase the UK’s productivity by 14.3% and grow GDP up to 10.3% by 2030.
So we know AI is the next big thing, but what’s the most effective way of utilising it?
Cloud technology has made accessing AI easier. In the past, you would need expensive infrastructure and specialist technology knowledge to use AI. Now, with the likes of Amazon Web Services and Microsoft Azure, you have the capability to spin up services instantly and assemble sophisticated human-like functionality. The barrier to entry is lower; people can get started without the need to understand the complex fundamentals.
This is already allowing a much larger set of hobbyists and organisations to experiment with AI than has previously been possible, but it’s not without challenges. Bolting on AI to existing situations without thinking about the quality of your data or overarching process can only provide limited benefits at best and unintended consequences at worst — and it certainly isn’t transformational.
One place we’ve encountered AI is within Robotic Process Automation (RPA) tools. So far RPA has had a mixed reputation with 30-50% of projects deemed to fail according to some analysis. A big part of that is due to its brittleness: for example, a software update rolls out and a new dialogue pops up stopping the bot dead in its tracks until an engineer can modify it. This is why the likes of Intelligent Automation are keen to infuse RPA with AI. Most RPA vendors are not currently using advanced AI but some are including things like image recognition and text analysis in their products. Perhaps this is just the beginning but RPA has a long way to grow if it is to become more than a tactical, sticking plaster solution. Currently, it doesn’t lead to true digital transformation. It doesn't streamline or cut the fat, and it can actually perpetuate legacy systems instead of digitising them.
To utilise AI to its fullest and be transformational, you need good quality data from your processes, which requires the whole end-to-end process to be digitised to capture that data in the first place. The rich data you collect from this digitisation will enable your AI to be more effective. If some of your processes are still manual, your data collection will be poor or incomplete and the AI will not get the whole picture. Poor quality and incomplete data will hinder your AI from making better decisions.
AI in government
A recent analysis by Deloitte identifies four distinct areas that AI can be applied within government.
- Relieve: Technology takes over mundane and repetitive tasks, freeing workers for more valuable work. For example, the US army has a chat bot that answers potential recruits’ questions.
- Split up: By breaking a job into parts, automation becomes more tractable even if there are some that continue to require human work. For example, HMRC uses RPA to create case numbers in a legacy system.
- Replace: In some cases the entire job can be automated but it’s important to recognise that for now these tend to be repetitive tasks with uniform inputs where decision-making that follows simple rules and tasks have a limited number of possible outcomes. Scanning post and routing for delivery is an example of this now long established.
- Augment: This is the true promise of AI: humans and computers combining their strengths to achieve faster and better results, often doing what humans simply couldn’t do before. Recent examples include the work on predicting which patients to recall for follow-up by doctors based on automated interpretation of test results.
In each case the common theme is that it’s essential to understand the work that needs to be done and the place each task has in achieving the mission of the organisation.
An organisation needs to understand and map its processes, streamline by cutting what’s unnecessary and then look at introducing automation, AI and other emerging technologies. AI in particular needs the contextual data you only obtain when you have your end-to-end processes digitised. This is a strategic and holistic approach Digital Process Automation takes, something we discussed recently in a public sector webinar.
You will then have a lean digital end-to-end process that AI can easily access data from. Remember, the more data AI has access to, the more accurate and efficient the AI will be.
Maximising the benefits of AI is a priority for government organisations. For citizens, it will result in a more personalised and efficient experience and for public sector workers, it means concentrating more time on innovative ways to improve services. Capitalising properly on this new technology will improve lives.