AI has made it easier than ever to submit complaints, FOIs and other previously friction-bound tasks.
asserted
it → make → complaints
Paul Ormerod considers how departments can wade through this new administrative slop
Revolutionary technologies always have unexpected side effects.
asserted
technologies → consider → effects
The release of ChatGPT just over four years ago marked the beginning of what we can think of as the truly modern AI era.
asserted
we → mark → era
In many ways it is hard to imagine that this was only in 2022.
asserted
this → imagine → 2022
No one anticipated at the time that generative AI would now be establishing important new results in maths.
asserted
AI → anticipate → maths
OpenAI has announced hundreds of proofs in the past few months, for example.
asserted
OpenAI → announce → example
This has sparked intense debate amongst mathematicians.
asserted
This → spark → mathematicians
The maths community has long accepted computer-assisted proofs which no human can check by hand, such as the famous 1976 result on the so-called Four Colour Theorem.
asserted
human → accept → Theorem
The argument is more about whether the proof advances human understanding.
asserted
proof → advance → understanding
A massive, opaque document tells mathematicians that a statement is true, but not why it is true.
asserted
it → tell → mathematicians
Still, some exceptionally hard problems do seem to have been solved.
asserted
problems → seem → ?
But here, too, generative AI has had unanticipated effects.
asserted
AI → have → effects
The cabinet secretary Antonia Romeo warned in the past week that a “surge of automated, high-volume citizen demand is hitting operational desks today” at an unprecedented “scale and pace”.
asserted
surge → warn → scale
AI agents are now submitting consultation responses on behalf of their users and drafting complex freedom of information (FOI) requests at the touch of a button.
asserted
agents → submit → button
New AI tools have been introduced in the civil service at least to try to deal with these tsunamis.
asserted
tools → introduce → tsunamis
They are explicitly designed to reduce the amount of time which humans need to spend on the submissions and complaints.
asserted
humans → design → submissions
But getting into an arms race with generative AI does not really feel like a permanent solution to the problem.
asserted
getting → get → problem
It is easier to think of potential solutions in two other, separate but related areas where AI is having a big impact.
asserted
AI → think → impact
The new threat to them is that the individual members themselves are discovering the generative AI can handle many of the issues which they have.
asserted
they → discover → which
The annual membership fee may no longer seem such good value.
uncertain
fee → seem → ?
The general secretary of the Prospect union has openly attacked this new trend.
asserted
secretary → attack → trend
Only last month Mike Clancy warned that “sycophantic” AI chatbots have led to a surge in claims from angry staff who wrongly believe they have a case for minor workplace grievances.
uncertain
they → warn → grievances
Unions, he said, were seeing a “spike in nonsense claims”.
uncertain
he → say → claims
In the same way, HR departments are being overwhelmed by complaints.
asserted
departments → overwhelm → complaints
Neil Morrison, the head of HR at Severn Trent, said those coming in are “quoting case law which is often not applicable”.
asserted
which → say → law
But they are lengthy and each one has to be given a proper response.
asserted
one → have → response
It is possible that this recent surge will fade away as people observe that demands for compensation are not succeeding in the way they imagined.
asserted
they → fade → way
Yet even if the reaction is entirely rational and staff do learn from the experience of others, this can take a long time.
asserted
this → learn → time
A techy problem needs a human response
The way forward lies with the HR departments readjusting their priorities.
asserted
departments → need → priorities
Instead of reacting to grievances, they should be in the forefront of efforts to generate a corporate culture in which complaints simply become less part of the staff ethos.
asserted
complaints → react → ethos
Social network research repeatedly shows that people are much more willing to accept unfavourable outcomes, for example, if they perceive the process to be fair.
asserted
process → show → example
AI changes the cost of complaining.
asserted
AI → change → complaining
It does not change human perceptions of fairness.
asserted
It → change → fairness
The way to deal with the problems created by AI is not by trying to have better technology, but to have better human responses.
asserted
way → deal → responses
You can follow him on Instagram @profpaulormerod
asserted
You → follow → @profpaulormerod