IN these fast-moving times, there’s no room for complacency. Just last week, the Prime Minister was in Berlin, hailing a new partnership between the UK and Germany to counter sabotage and cyber-attacks amid growing concern about “hybrid attacks” from Russia combining military and non-military methods.
“Attacks on our infrastructure, our businesses and our democracies are not a distant threat,” said Andy Burnham. “They reach into people’s homes and workplaces, and they are designed to make us weaker and more divided.”
It’s reassuring to know Europe’s two biggest economies are working together on this, but to read Monday morning’s headlines, one might ask if the UK – or any other state – actually needs enemies when its own citizens (and even its own politicians) are so willing to cause disruption and paralysis by using generative AI.
Ask a government worker, or the manager
Ask a government worker, or the manager of a GP practice, and they might tell you that it feels like the cyber-attack is coming from inside the house.
The Scotsman reports that civil servants are at their wits’ end trying to deal with a deluge of parliamentary questions and freedom-of-information requests that they believe are being generated using artificial intelligence.
The number of demands for information from certain MSPs has been raising suspicions for some time, in particular Tory MSP Stephen Kerr’s 1477 questions in one year and his colleague Douglas Lumsden’s even more staggering 987 in a single month.
It was not just the volume of those
It was not just the volume of those questions that raised eyebrows but their nature, relating as they did to such vital matters as potato peelers and salt and pepper sachets. The Tories’ claims that they are merely holding the Scottish Government to account over its spending are hard to take seriously when the average cost of answering each of these questions is almost £140.
Every minute answering these questions and preparing briefings is time civil servants could and should be spending on actually delivering policies. Is sending them off on wild goose chases any less of a threat to the functioning of the country than, say, infecting their laptops with viruses that prevent them from working for the equivalent number of hours? It certainly feels more like an act of sabotage than a process of legitimate scrutiny.
The ability to produce large volumes of text with minimal effort is also being exploited by patients writing complaints to their GP practices, with more than 90% of UK practice managers saying they believe they have had AI-generated complaints, some of which run to 20 pages or more and contain irrelevant information and legalese.
Complaints cannot be ignored, but the length
Complaints cannot be ignored, but the length and form of these screeds make them difficult to decipher and investigate. The Medical Protection Society, which provides advice and support to healthcare workers, notes that the ease with which they can be produced is “lowering the thresholds around what patients will complain about, and encouraging more complaints to be made in the heat of the moment”. It hardly takes much wit or time to fire off a quick email with a bullet-point summary of a bad experience. Trying to enter into correspondence with someone illiterate is only likely to produce further incoherent text.
At a time when access to GP services is a major ongoing challenge, and A&E departments are overrun with people who don’t need to be there, it’s utterly frustrating to contemplate doctors and nurses having to spend any time fielding queries from admin staff about spurious or incomprehensible complaints. As it happens, one of the civil servants quoted in The Scotsman’s story said their efforts to progress the Scottish Government’s GP walk-in centres policy – designed to alleviate some of the pressures – is being hindered by demands for briefing notes about the policy.
Artificial intelligence may yet unlock an array of new treatments and cures, but if the people in a position to approve and administer them are drowning in digital paperwork, patients will be no further forward.
A final gloomy headline in The Times describes
A final gloomy headline in The Times describes how authors are being duped into believing their work has been discovered by someone who wishes to publish or promote it. Of course, it has not been read by a human being, but fed into AI. The authors targeted are not daft, but these days one need not be to take a scam at face value.
At an individual level, none of the AI use described above might sound like a major threat to our infrastructure, our businesses and our democracies, but cumulatively, it’s not hard to see how this kind of behaviour, carried out at scale by individuals with their own agendas, could cause massive disruption. If the answer is to fight fire with fire – ie, use AI to respond to questions, complaints or opportunities – the outcome could well be even worse.
In all aspects of society, the use of generative AI is eroding trust and – to return to the PM’s words – making us weaker and more divided. We are doing this to ourselves, and each other. Is there a way back?
