AI Software in 911 Dispatch Centers: An Innovative Solution

While the use of 911 is commonplace today, this life-saving universal number is actually less than 80 years old. The first concept of 911 dates back to 1957, as a result of the National Association of Fire Chiefs rallying for a standardized phone number that the public could call to report fires. Over time, this evolved into a number for reporting emergency situations in general, not just fire-related incidents. In 1967, the Federal Communications Commission began collaborating with the American Telephone & Telegraph Company (AT&T) on the creation of a number for the public to use. The numbers 9-1-1 were chosen because it was a short number sequence, easily remembered, and just as easily dialed. Additionally, that number sequence was not currently in use as an area code or service code, which made the implementation of the number less cumbersome to operationalize.

911 Types

Since that time, the nation’s 911 system has evolved into multiple types identified as Public Safety Answering Points (PSAPs), or emergency call centers. There are three main subsets of the 911 system:

• Basic 911: A person takes the call from the PSAP, and all communication occurs by voice between caller and call-taker.
• Enhanced 911 (e911): The call is routed to the proper PSAP based on the caller’s location. Additionally, the caller’s phone number and address are displayed to the call-taker. Most counties with 911 coverage have converted to e911 (also called “wireless 911”), which has two phases:
  • Wireless Phase I: The call-taker automatically receives the caller’s number and the location of the cell tower handling the call.
  • Wireless Phase II: The call-taker automatically receives both the phone number and location of the caller.
• 911 Voice over Internet Protocol (VoIP): VoIP simply means that the phone call is being made over an internet connection instead of a traditional phone line.

Evolution of the System

Several years after the first 911 call was made, the White House encouraged nationwide implementation of 911. From the 1980s to the late 1990s, the percentage of the population being serviced by 911 moved from 50% to greater than 90%. As of 2024, the U.S. has 8,865 primary and secondary PSAPs. The National Emergency Number Association also estimates that 240 million 911 calls are made each year as of 2021, and more than 80% of the calls are wireless. Additionally, Next Generation 911 (NG911) is a digital internet protocol system created to replace the legacy analog structure and to secure the resiliency of 911.

911 Staffing and Shortages

Minimum training expectations are not currently standardized in the U.S. for PSAP telecommunicators (TCOs). Education and qualifications across these centers range from a high school diploma or GED requirement with on-site training, to requirements for first responder or emergency medical technician experience. Due to the lack of standardization of requirements, some PSAPs have a larger burden in terms of training new TCOs, particularly those without any medical training. Additionally, emergency dispatch has a high turnover and burnout rate by the nature of the work. Coupled with staffing challenges, mental health and work-life balance difficulties in this profession are cause for concern, as many TCOs work longer shifts and overtime to help bridge critical gaps.

Staffing shortages are often filled by supervisors, which takes away from their other duties, such as quality assurance, real-time (and continual) feedback and training of new TCOs, and feedback and development for seasoned staff. This can sometimes create a vicious cycle, leading to additional turnover and staff shortages. A 2023 report suggests that PSAP staffing shortages created, on average, a 25% vacancy rate, with some centers as high as 70%. These shortages are a pervasive and continuing issue across all states. The report also identified that turnover is a problem among new hires and seasoned TCOs, citing reasons such as pay, work hours, and increased call volume, further supporting the concern about mental health, burnout, and work-life balance in this profession.

By allowing AI to take some of the burden from existing staff and leadership, focus can be returned to where it is most needed within each PSAP. Incorporating artificial intelligence (AI) into some, if not all, aspects of the PSAP process can be cost-effective. The average cost to implement AI can be comparable to money spent on part-time salaries or as-needed staff, TCO, or on overtime paid due to staffing shortages.

Ways that PSAPs Can Use AI

AI software is meant to serve as a support system, rather than a staff replacement, and its use is primarily to enhance the overall effectiveness of operations. The number of relevant AI programs available to the industry continues to increase, offering many work options based on the needs of each organization or PSAP. Several programs combine the use of call audio, radio communications, and computer-aided dispatch to develop content to analyze data and identify trends. PSAPs can use these programs in some or all areas of operation, such as the following:

• Call triage: Streamlining call flows and identifying the nature of the emergency
• Call routing: Moving the call to the correct agency (law enforcement, fire, medical)
• Call diversion: Moving non-urgent calls into the correct workflow, which also opens an opportunity for AI to assist with non-emergency calls
• Automated callback: Streamlining responses based on accidental dialing versus needed dispatch intervention
• Detection of call surges: Routing call increases from an area that already has response enroute to free TCOs for more emergent calls
• Geofencing call hotspots: Identifying areas of most call volume in the event of a mass casualty to prioritize the dispatch of responders and resources
• Note-taking: Maximizing documentation efficiency by reducing the risk of missing items and slow typing speeds
• Translation and transcription:
  • Providing immediate, multi-language support in a culturally diverse environment
  • Deciphering difficult-to-understand speech patterns into words
  • Real-time transcription, minimizing the time a TCO needs to ask the caller to repeat themselves
  • Patient analysis and suggested relevant questions to ask the caller
• Data analysis: Anticipating necessary resources and interventions based on real data within the previous call volume

Using AI for Quality Assurance

AI is also useful as a quality assurance mechanism for PSAPs. Quality assurance is traditionally the responsibility of a supervisor, who pulls a random percentage of calls for each TCO and evaluates a variety of systems, including call audio, radio recordings, and TCO documentation. Many AI programs can combine these tasks and analyze all calls that come into the PSAP, which can help identify areas of focus that may be missed in a limited random sample. Supervisors can then review the results of the AI analysis to determine if additional training is warranted.

Some AI programs offer complete analysis of the TCO, helping to identify readiness post-training. Additionally, these components can be monitored over time, identifying competency or room for improvement based on the individual. Examples of skills AI can assess include call-taking, typing speed, decision-making, and multitasking.

AI programs can also be used to evaluate the overall team, a shift, or an individual. This helps identify larger training gaps or educational opportunities versus programs that require individual mentoring and attention. Larger gaps can be used to create tabletop exercises and strengthen areas of concern.

Most AI dispatch programs provide real-time data on every received call, which can provide a greater focus on reviewing individual TCOs’ areas of opportunity and strengths. Several programs offer pre-populated forms that can be used to review with personnel. Calls can be separated by line of work, such as law enforcement vs. fire; basic call vs. advanced call. Specific criteria can customize programs to offer standardized, formal performance reviews for tracking and analysis.

AI-enabled programs also can delve into specific concentrations such as policy and protocol, communication processes, and overall accuracy. They can identify superior performance based on preset criteria, such as clarity of communication, providing exceptional instructions or caller gratitude. This is traditionally identified through call analysis, direct observation, or colleague referral. The ability to gather more data with AI assistance minimizes the potential to overlook exceptional performance that should be recognized. With more deserved recognition, staff would feel more valued, which could increase retention rates. Conversely, some software tracks emotion or tone of voice, alerting management to potential burnout.

Areas of Concern or Unintended Consequences

As with implementing any new system or process, there will be risks and concerns to address if adopting AI assistance at a PSAP.  One of the more obvious concerns would be gaining public trust and acceptance of the process.  In our strongest time of need, most people want to hear the sound of another human being when they call for help.  Strong monitoring of any AI logic solutions related to re-routing or triaging calls is critical for the success of implementing AI, as well as gaining and maintaining public trust of the process.  Significant testing is recommended prior to rolling out either of these approaches.  It is important to communicate that AI is an enhancement / aid, not a replacement for the TCOs within a PSAP.

Additionally, digital operations increase the risk of cyberattacks and the overall vulnerability of the 911 system. While these attacks are becoming more frequent and are difficult to prevent, it is important to regularly assess risk and form solutions where possible. Increased protections will need to be put in place to minimize unintended consequences of digital solutions to current problems.  While cyberattacks are an increasing risk, AI can also be used to combat swatting techniques that are also becoming a similarly frequent issue. Another concern that warrants additional research relates to mental health effects. If AI re-routes or takes over less emergent calls, TCOs will necessarily take more problematic calls than previously, which may increase individual stress. When implementing AI, TCOs should be monitored for any mental health concerns that need to be addressed as a result of this change.

Each county should assess the pros and cons of implementing AI within their PSAPs to help streamline operations, decrease administrative burdens to leadership, increase efficiencies, and combat continued staffing issues experienced within these centers. A financial review of staffing and overtime expenses compared against the cost to implement AI in any part of the dispatch process could be a strong solution against turnover and burnout.