AI To Triage UK Police 101 Calls
UK police are beginning to use artificial intelligence to decide where non-emergency 101 calls should go before they reach a human operator, with a £1.4 million system designed to cut waiting times, reduce wasted police time and redirect millions of misdirected calls more efficiently.
What Is Changing?
The new Natural Language Processing system has already gone live with South Yorkshire Police after trials involving 15 forces.
Rather than placing every caller into the same queue, the AI listens to why somebody is calling and matches the request against likely reasons. If the issue is not a police matter, the caller can be redirected to the appropriate organisation before speaking to a police call handler.
That could include local councils, NHS 111 or the 105 electricity power-cut helpline.
The scale of the problem helps explain the investment, with the 101 service receiving around 20 million calls nationally each year. It seems that approximately four million of them, or 20 per cent, are actually intended for organisations other than the police. In fact, some calls are simply inappropriate, with examples including complaints about delayed pizza deliveries, slow pub service and requests for lifts, all of which take up time that could otherwise be spent helping people with genuine policing matters.
Why Use AI To Help?
At present, a misdirected caller can still consume a call handler’s time before anyone establishes that the police can’t help. Therefore, automating that first stage should allow trained staff to spend more time dealing with crime reports, victims and safeguarding concerns.
Superintendent Richard Hammond, Operations Manager at South Yorkshire Police’s Force Control Room, explains: “This new NLP software will reduce unnecessary calls coming through to our Force Control Room, but everyone who does need to speak to the police will be put through to a human call handler.”
Savings Justify The Spend
The government estimates the system could actually save policing up to £8.5 million a year, giving the £1.4 million investment a potentially strong return if it performs as expected.
Costs aside, the bigger objective here appears to be to prevent administrative demand from delaying people who genuinely need police assistance.
How Does The AI Know Where A Call Should Go?
The system uses natural-language processing, or NLP, which allows software to analyse spoken language and identify what somebody is trying to achieve.
For example, instead of expecting callers to navigate a rigid menu, it can interpret ordinary descriptions of a problem and categorise the reason for the call.
That matters because accents, vocabulary and phrasing vary considerably, so South Yorkshire Police says the software has undergone extensive training and testing intended to ensure its performance is not affected by those differences.
Thankfully, as anyone who regularly uses AI will be hoping, there are also human safeguards in place. For example, if somebody genuinely needs police help, they should still be able to reach a human call handler, while uncertain situations can be escalated rather than automatically diverted.
As Superintendent Richard Hammond says: “The software will filter out calls that are not of a policing nature before they come through to our call handlers. These callers will be directed to the most appropriate agency that can offer them the assistance they need.”
Part Of A Wider Effort
The system is actually part of a wider effort to modernise public contact with policing.
The government is investing £16.5 million in digital public-contact improvements, including AI transcription for 999 and 101 calls and tools designed to link crime reports to identify patterns in demand.
That suggests the important change is not simply one AI call-routing system. Police control rooms are gradually becoming environments where AI performs more repetitive sorting, transcription and analysis around human decision-making.
Crime and Policing Minister Sarah Jones said: “Every minute spent on these calls is time taken away from helping people who genuinely need their support.”
She added that AI could help “bring control rooms into the 21st century, freeing up time and unlocking millions of pounds”.
What Could Possibly Go Wrong?
Using AI at the first point of public contact inevitably raises questions about accuracy. Misclassifying a routine enquiry could cause inconvenience, but misunderstanding somebody trying to report a crime, vulnerability or safeguarding concern could have much more serious consequences.
That makes the fallback arrangements in this case extra important. It’s worth remembering here that the aim isn’t to replace human judgement in difficult cases, but to stop clearly inappropriate calls reaching people whose time is better spent elsewhere.
Its success will, however, depend on how reliably it recognises intent and how readily uncertain calls are passed to humans.
What Does This Mean For Your Business?
The most useful lesson for businesses here is that AI doesn’t have to replace an employee to produce some worthwhile savings or improve how things run.
Many organisations receive large volumes of calls, emails, support tickets and enquiries that first need to be understood and routed. At this stage in its development, AI can potentially handle that initial sorting, directing straightforward requests to the correct team while leaving employees to concentrate on cases requiring judgement, expertise or empathy.
The police example also shows where some caution is needed. Automation works best when the boundaries are very clear, escalation is easy and humans remain available when the system is unsure.
For businesses looking for practical AI projects, that may be a better starting point than trying to automate entire jobs. The immediate value may come from removing repetitive work around skilled employees, rather than trying to remove the employees themselves.
