Image © iStock
Speaking in a session entitled Balancing the risks and rewards of AI in public services, researcher Dr Stephanie Hare gave an update on the state of play on AI, which in the past 18 months has dominated the technology agenda, with ChatGPT style gen-AI creating a buzz around the possibilities.
Understanding the tech is key to its appropriate use, Dr Hare explained. She also outlined some of the emerging risks that target AI-based tools – from data poisoning, where malefactors tamper with large datasets can, to prompt injection threats which involve hacking LLMs to alter the directional algorithm, AI is not immune from the cyber threats that plague other forms of technology.
“There are real risks to this technology,” she explained. “This is a technology in Beta – it’s still in development, so you put it into your organisation at your peril. You cannot completely trust it, because it’s inaccurate and it invents things. “
So what should public sector organisations do to ensure they stay abreast of the latest developments in AI but not to a point where they’re exposed to excessive risk? Dr Hare called on finance to step up to the plate to make sure AI is used responsibly: “I’m a fan of AI, but I want it to be used responsibly. So AI that helps a radiologist read a CT scan better and more accurately? That sounds great.
“However, I’m much more cautious about how the patient feels about it, and I want to see doctors and nurses well trained in it. So you can’t just impose this on people – you have to focus on transparency for patients. You have to take them on that journey.”
Dr Hare made it clear that one of finance’s most important functions in the new age of AI was to bring in a dose of healthy scepticism and rigour to the roll out of AI in the public sector. “Be specific – look at the types of problems that your organisation is trying to solve. The chances are that others are working on it so it’s worth seeing what they’re doing and learning from that.”
Meanwhile Dr Raphael Olaiya, lead data scientist at the NHS, agreed that the challenges facing the health service in particular have to be faced up to, and solutions found. Whether that’s long waiting lists and a lack of staff, with many overworked and suffering low morale, to a stubborn focus on disease rather than on wellbeing and quality of life, Dr Olaiya suggested that the key barrier to improvement wasn’t a lack of funds, but rather irresponsible expenditure, where ‘bad spending’ – a problem that AI can address.
He suggested the skillsets needed to use AI effectively were beginning to emerge in the NHS and that the next five-to-10 years will see a “tipping point” where we enter the era of ‘Healthcare 2.0’ the number of public sector staff with the right training are in place.
But he sounded a note of caution: “We need to make sure our vision is right. The technology is going to get there, but are we ready to use it effectively? At the end of the day, the aim of healthcare is to treat disease, so there is a cultural shift that needs to happen to get people on board with this. And training has to be done well.”
The benefits of AI-driven tools in public sector organisations were outlined in a later session, supported by Kefron Digital. Elaine Simpson, head of finance service centre at East Midlands Shared Services, took delegates through the rollout of an AI-enabled tool to help the centre improve its operations. “The centralisation helps give you better visibility over your processes and reporting, and having that allows you to decide on the areas you want to focus on across the organisation. Having that info there is key.
“And then it’s about accountability – people now know where the lines are, who’s responsible for what and you can see if it breaks down. And then finally we’ve gone touch free, so if a purchased order has been receipted, then when the invoice comes in, it automatically scans and captures the data and matches it against the payment – so we don’t usually have to touch it. and that’s freed our time up to work on more valuable projects.”
