Techno Blender
Digitally Yours.

AI for 911 operators and the future of emergency services

0 45



In the course of a few months, generative AI has become ubiquitous: fighting forest fires, caring for seniors, predicting heart disease, and developing cancer drugs. OpenAI’s ChatGPT, whose meteoric debut quickly made it the technology’s best-known example, reached 100 million monthly active users within two months—a record time. Microsoft and Google have announced new, AI-powered features for their workplace tool suites, Microsoft 365 and Google Workspace, respectively, which will transcribe, take notes, write reports, and compose emails. Google released its own chatbot, Bard. Zoom is incorporating Anthropic’s chatbot into its teleconference platform. Meta has pivoted to its LLaMA AI while Stability has produced the first open-source chatbot, StableVicuna. 

That buzz has also overshadowed a quieter AI revolution, where emergency responders have incorporated AI into their systems and processes. If you’re not ready to trust your life to Siri or Alexa during an emergency, don’t worry, neither is the technology—yet.

But the way AI already helps 911 call-takers gives a glimpse of how the technology may save lives in the future. And yes, humans will keep answering your 911 calls despite the fact that a shortage of workers is a major factor driving this transformation.

We place roughly 240 million 911 calls annually, which averages out to more than 650,000 per day, though studies have shown that most of these calls are not emergency-related. At the same time, police departments are enduring staffing shortfalls which the pandemic’s ebb has not abatedespecially among 911 call-takers, who often endure long hours in high-stress roles. This brain drain is pulling critical experience out of call centers, meaning that those taking the calls are newer to the job. Increasing urban diversity means that fewer 911 callers are native English speakers, only complicating matters further. These factors are already causing mistakes and delays which will only worsen.

Enter assistive AI, whose purpose is to help humans work more efficiently by automating rote and simple tasks while making overwhelming ones like large-scale data analysis, for example, more manageable. Departments across the country are already leveraging the technology to ease their telecommunicators’ non-emergency burden by using AI to triage non-emergency calls. Cities like New Orleans, Austin, Portland, and San Jose are deploying, or planning to deploy, AI to help deal with issues raised through their 311 lines, such as pot-holes and cats stuck in trees.

Many departments also use AI-powered transcription to support their 911 call-takers, the same type of technology which permits you to transcribe your Zoom or Google Meet gatherings. In effect, AI can act as a digital ear capable of listening to and writing down spoken conversation. Not only does it produce a real-time, verbatim record of the call but it can also flag key details such as location and the nature of the emergency. This allows the call-taker to focus on the emergency and manage the best response rather than note-taking and documentation.

The next steps involve context, which is not something of which computers have an innate understanding. Neither is experience—how to connect past events with outcomes to predict what will or should happen next.

Interestingly, though, large language models, such as ChatGPT and Bard, do retain information in a semantically related way where language, concepts, and past outcomes are connected. These models, in other words, are learning, in this case, to use experience as a guide. 

This opens opportunities, especially to mitigate the experience gap. Imagine these near-future scenarios: 

  • AI can help resolve the growing language barrier. Between 1980 and 2019, the number of people in the U.S. who speak a language other than English at home nearly tripled. Using Massively Multilingual Speech models, AI software will be able to detect and operate in numerous languages, including those for which there is little or no data. Further ahead, but closer than you might think, will be real-time translation, akin to Star Trek’s universal translators.
  • AI will be able to analyze previously-transcribed calls and situations in order to present to the call-taker various suggested options and responses to the current situation. This will have the salutary effect of harnessing collective experience even as the number of veteran call-takers continues to decline. The AI could become a virtual institutional memory available to every call-taker from day one.
  • As the call-taker secures critical information about the unfolding emergency, the AI will be able to simultaneously create a new workflow around it, ensuring that the other relevant law enforcement entities—dispatch for example—are looped in. This will ensure smoother and more rapid mobilization of resources to handle the situation.
  • AI will be able to review all calls and incidents for quality-assurance purposes. Most emergency response systems review a randomly selected 2% of all incidents but generative AI will be able to do 50 times the work in a fraction of the time. This will not only provide better and more comprehensive feedback but will also help leadership better plan staffing schedules, etc.

Iterative AIs must still surmount one major emergency response obstacle: testability. The very flexibility which makes these programs so powerful also makes them hard to thoroughly assess. Every call is unique, so every answer will be, too. How do you know when they’re ready for a specific job, especially one with the stakes of a 911 emergency? Offering the right suggestions and correct information most of the time is not good enough when lives could be on the line.

Anticipating every contingency is impossible, so extensive real-world testing will be necessary, but also not feasible given the stakes of a 911 call. That’s why using AI to help humans with non-emergency calls is such an important start, because it will allow us to better understand and ameliorate its shortcomings without a mistake having significant consequences.

When AI reaches the stage where it can be a full partner for 911 operators, it will probably not be a ChatGPT viral moment, but it could be something more important: a lifesaver.


Mahesh Saptharishi, PhD, is the executive vice president and chief technology officer at Motorola Solutions.





In the course of a few months, generative AI has become ubiquitous: fighting forest fires, caring for seniors, predicting heart disease, and developing cancer drugs. OpenAI’s ChatGPT, whose meteoric debut quickly made it the technology’s best-known example, reached 100 million monthly active users within two months—a record time. Microsoft and Google have announced new, AI-powered features for their workplace tool suites, Microsoft 365 and Google Workspace, respectively, which will transcribe, take notes, write reports, and compose emails. Google released its own chatbot, Bard. Zoom is incorporating Anthropic’s chatbot into its teleconference platform. Meta has pivoted to its LLaMA AI while Stability has produced the first open-source chatbot, StableVicuna. 

That buzz has also overshadowed a quieter AI revolution, where emergency responders have incorporated AI into their systems and processes. If you’re not ready to trust your life to Siri or Alexa during an emergency, don’t worry, neither is the technology—yet.

But the way AI already helps 911 call-takers gives a glimpse of how the technology may save lives in the future. And yes, humans will keep answering your 911 calls despite the fact that a shortage of workers is a major factor driving this transformation.

We place roughly 240 million 911 calls annually, which averages out to more than 650,000 per day, though studies have shown that most of these calls are not emergency-related. At the same time, police departments are enduring staffing shortfalls which the pandemic’s ebb has not abatedespecially among 911 call-takers, who often endure long hours in high-stress roles. This brain drain is pulling critical experience out of call centers, meaning that those taking the calls are newer to the job. Increasing urban diversity means that fewer 911 callers are native English speakers, only complicating matters further. These factors are already causing mistakes and delays which will only worsen.

Enter assistive AI, whose purpose is to help humans work more efficiently by automating rote and simple tasks while making overwhelming ones like large-scale data analysis, for example, more manageable. Departments across the country are already leveraging the technology to ease their telecommunicators’ non-emergency burden by using AI to triage non-emergency calls. Cities like New Orleans, Austin, Portland, and San Jose are deploying, or planning to deploy, AI to help deal with issues raised through their 311 lines, such as pot-holes and cats stuck in trees.

Many departments also use AI-powered transcription to support their 911 call-takers, the same type of technology which permits you to transcribe your Zoom or Google Meet gatherings. In effect, AI can act as a digital ear capable of listening to and writing down spoken conversation. Not only does it produce a real-time, verbatim record of the call but it can also flag key details such as location and the nature of the emergency. This allows the call-taker to focus on the emergency and manage the best response rather than note-taking and documentation.

The next steps involve context, which is not something of which computers have an innate understanding. Neither is experience—how to connect past events with outcomes to predict what will or should happen next.

Interestingly, though, large language models, such as ChatGPT and Bard, do retain information in a semantically related way where language, concepts, and past outcomes are connected. These models, in other words, are learning, in this case, to use experience as a guide. 

This opens opportunities, especially to mitigate the experience gap. Imagine these near-future scenarios: 

  • AI can help resolve the growing language barrier. Between 1980 and 2019, the number of people in the U.S. who speak a language other than English at home nearly tripled. Using Massively Multilingual Speech models, AI software will be able to detect and operate in numerous languages, including those for which there is little or no data. Further ahead, but closer than you might think, will be real-time translation, akin to Star Trek’s universal translators.
  • AI will be able to analyze previously-transcribed calls and situations in order to present to the call-taker various suggested options and responses to the current situation. This will have the salutary effect of harnessing collective experience even as the number of veteran call-takers continues to decline. The AI could become a virtual institutional memory available to every call-taker from day one.
  • As the call-taker secures critical information about the unfolding emergency, the AI will be able to simultaneously create a new workflow around it, ensuring that the other relevant law enforcement entities—dispatch for example—are looped in. This will ensure smoother and more rapid mobilization of resources to handle the situation.
  • AI will be able to review all calls and incidents for quality-assurance purposes. Most emergency response systems review a randomly selected 2% of all incidents but generative AI will be able to do 50 times the work in a fraction of the time. This will not only provide better and more comprehensive feedback but will also help leadership better plan staffing schedules, etc.

Iterative AIs must still surmount one major emergency response obstacle: testability. The very flexibility which makes these programs so powerful also makes them hard to thoroughly assess. Every call is unique, so every answer will be, too. How do you know when they’re ready for a specific job, especially one with the stakes of a 911 emergency? Offering the right suggestions and correct information most of the time is not good enough when lives could be on the line.

Anticipating every contingency is impossible, so extensive real-world testing will be necessary, but also not feasible given the stakes of a 911 call. That’s why using AI to help humans with non-emergency calls is such an important start, because it will allow us to better understand and ameliorate its shortcomings without a mistake having significant consequences.

When AI reaches the stage where it can be a full partner for 911 operators, it will probably not be a ChatGPT viral moment, but it could be something more important: a lifesaver.


Mahesh Saptharishi, PhD, is the executive vice president and chief technology officer at Motorola Solutions.

FOLLOW US ON GOOGLE NEWS

Read original article here

Denial of responsibility! Techno Blender is an automatic aggregator of the all world’s media. In each content, the hyperlink to the primary source is specified. All trademarks belong to their rightful owners, all materials to their authors. If you are the owner of the content and do not want us to publish your materials, please contact us by email – [email protected]. The content will be deleted within 24 hours.

Leave a comment