Posted on: 10/05/2021
As published in the Emergency Services Times.
The majority of emergency responses start with a phone call, and capturing those conversations has always been critical to assist with operations, for investigative purposes and to meet legal obligations. For dispatch workers on the front line at risk of stress-related mental health issues resulting from higher call volumes due to the pandemic and reduced resource, recordings can provide insights that can help team leaders understand agent experience and welfare.
Today, advances in machine learning and artificial intelligence – coupled with a massive uptick in computing power – means voice data is now processable at scale, significantly expanding the use cases and value of voice data.
We know voice-based interactions remain a key communication channel. Even though NHS England created a dedicated online resource to support sharing information with the public during the pandemic, it still received over 95,000 calls a day throughout March 2020, an increase of over 6000% when compared with the year before. While this automation is impressive, it shows that people will always rather talk to agents where complex interactions and issues are concerned. It is these conversations which are particularly rich in terms of the insights they can deliver.
Voice is an especially significant data point. It can convey intent, sentiment, emotion, action and context. This is all information that is often difficult to discern in other forms of data. For those in the control room, a voice call is a critical piece of the investigative puzzle. Speed, accuracy and reliability are prerequisites of any solution designed for the emergency services and an effective voice solution can improve call handling and enable a controller and investigator to verify important details that may be critical to an emergency response.
Control room operators, by their nature, are required to work at a high technical level, perpetually under time pressure, and are dealing with calls that are quite often a matter of life and death. Operational intelligence and analytics, coupled with understanding conversations across control rooms and all touch points to provide a holistic view, are key. Organisations are now proactively looking to move to data-driven processes and decision making, ultimately through digital transformation, to address these pressures.
A survey we conducted during a presentation at the recent BAPCO Online event highlights this further. Though more traditional business cases for voice are still prevalent, with 45% of respondents citing compliance and Quality Management as key use-cases, we are also starting to see AI and analytics growing in importance (18%). In addition, agent welfare, workforce optimisation and customer experience initiatives are also growing in priority, demonstrating the wide opportunities voice presents within the control room.
There are many challenges presented by a significant increase in call volumes and indeed the ongoing issues with inappropriate or hoax calls such as members of the public wanting to order takeaway food, or repeat callers suffering from mental health issues who call several times a day for ‘a chat’.
Using technology to triage calls is one way that artificial intelligence (AI) can assist in a control room. However, use of such new technology is not without dangers and mishaps if it is not introduced correctly.
A flagship artificial intelligence system designed to predict gun and knife violence, known as Most Serious Violence (MSV), and part of the National Data Analytics Solution (NDAS) project funded by the Home Office, was withdrawn in its current form due to several failures. One study, published by RUSI, called for the introduction of national guidance to oversee the police’s use of data-driven technology amid concerns that it could lead to discrimination. It argued that officers are battling against ‘information overload’ as the volume of data around their work grows, while there is also a perceived need to take a ‘preventative’ rather than ‘reactive’ stance to policing.
Data-driven technology can help keep the public safe if it is developed in conjunction with the emergency services, ensuring guidelines are met and building confidence in its ongoing use.
As mentioned earlier, we saw a significant rise in call volumes throughout the pandemic which, while a logistical challenge, has provided even more data to analyse and use for improving the service. This abundance of call recording data and interactions provides a wealth of information that can be fed into AI engines to deliver timely, actionable insights from the analysis of data at scale.
With 40% of local authority and public sector organisations expecting speech analytics to become a strategic asset within under two years (research from Sapio, November 2020), it’s critical that organisations consider data access, data security and data portability as part of their strategy.
Working with vendors with an open API philosophy is a key element of successfully leveraging voice data within AI and analytics applications. By retaining control of captured data, organisations can choose best of breed vendors that provide the capabilities that best meet their requirements, rather than being locked into a single application stack. Our own research suggests that 51% of an organisation’s captured conversations are inaccessible for AI and analytics. Far from signing up to a siloed approach, it is essential for organisations to expect their data capture solution to be always on, always open, always accessible.
We are working with the UK Government and its partners to ensure organisations are fully supported in their migration to DCS and the Emergency Services Network (ESN) as well as ensuring they are enabled to leverage AI tools and applications in the future.
The replacement of legacy technology, coupled with providing a common platform for innovation and data sharing, further emboldens the technological capability of what is possible. With the right technology partners providing the tools, organisations can ensure they are ready for future spikes in call volumes, the migration to DCS and ESN, and are fully prepared and enabled to leverage AI when they are ready, whether that’s for predictive policing, crime prevention, trend spotting, agent welfare, or whatever the use case.