In brief

In brief

  • Deep learning, natural language processing and computer vision could improve public safety in countless ways.
  • Quick wins include automaton for case management and equipping law enforcement with algorithms for real time analysis.
  • To prepare, agencies should organize their data, embrace the cloud and invest in needed training.

Accenture: What will be the impact of Artificial Intelligence on public safety?

Nilanjan Sengupta: AI’s potential across public safety is significant. Whether applied to law enforcement, first responder or homeland security mission areas, there are many compelling use cases around threat detection, incident response, case management, citizen service, cyber security and streamlined operations which can all benefit from AI.

The opportunity and the challenge for many agencies is the vast amounts of digital information being created today. In 2013, the FBI needed to quickly assess more than 50 terabytes of data to investigate the Boston Marathon bombing. In contrast, for 2017’s Las Vegas shooting, the amount of data had grown twentyfold, to more than a petabyte. AI can help us manage this onslaught of data to find important answers faster and more efficiently.

Over the past several years, AI’s ability to solve these problems has improved dramatically, as has the accessibility of the technology. We have made significant strides in computer vision and image detection, applying deep learning, neural networks, convolutional and recurrent neural networks, and things of that nature to identify individuals and actions by scanning and looking at images and videos. We’ve seen similar advances in natural language processing—understanding real—time speech and text to potentially identify particular threats. This positions AI as an emerging core competency for public safety agencies.

A: How can public sector begin to prepare for coming AI applications?

NS: Data is the fuel that powers AI. This means that we need an integrated data environment, one that is organized around mission outcomes or use cases that we are trying to support. In many cases, we need to combine information from multiple sources and data types in a secure way to enable the models and analysis that we want to create.

For many agencies, these efforts are already part of their digital transformation strategy, as AI is a fundamental component of this next-generation operating model. The real challenge is embedding these capabilities into the enterprise fabric so employees can tap into the power of AI natively.

The needed skills surrounding deep learning, computer vison and other aspects of AI are the other critical element of readiness. This is important as AI is such an evolutionary field requiring a lot of experimentation to develop effective solutions. Beyond data scientists, we also need skilled cloud engineers that can build the agile, scalable infrastructure needed to process all of this data.

A: How will all this impact the human side of public safety?

NS: The way we consume information today, it’s very reactive: We see what just happened. AI is going to give us forward-looking information; it will be predictive. People will need to learn how to look at that information and how to act on that information. Training will play a tremendous role in all of this.

Then for public safety in particular we look at the people at the edge—people driving a police car or an FBI vehicle or a border patrol mission. They need to be equipped not only with gear and uniforms but also with AI and algorithms so that they can detect different things that are going around them. We will need to train people to operate in that environment.

Integrating this insight into existing workflows in a productive manner can be challenging. We don’t want frontline officers distracted by false alerts and extraneous information. This is one of the reasons that we focus so much on human-centered design as we want to ensure that we are providing them with the situational awareness that they need.

A: What are some quick wins that are most readily available to public safety?

NS: Using automation to help manage cases and clear backlogs is a compelling focus for many agencies. Similar to other aspects of government, elements of AI can be used to perform repetitive tasks, allowing professionals to spend more time on deeper case work and analysis. The benefits include faster processing and adjudication, improved accuracy, and potential cost-savings.

A: How important is the cloud to taking advantage of AI?

NS: The cloud is a significant force multiplier for AI. Critical in many cases as leading AI capabilities are often cloud-native. With the cloud, it is also easier to create an integrated data environment and extensible data science platform. This streamlines access to data, delivers critical throughput and elasticity, improves collaboration, and makes it easier to add new capabilities.

A: How is Accenture innovating with AI in public safety?

NS: Through the Accenture Federal Digital Studio, we have worked with a number of public safety agencies to reimagine their operations. The studio’s AI Lab is helping us identify new ways to use data to improve public safety, and rapid prototyping capabilities are creating new solutions faster. We also work with clients to embed these capabilities within their agencies through centers of excellences and fusion centers.

Accenture’s size and capability breadth also give us unique insight into the rapidly evolving technology landscape. We are often the largest partner for many of the most strategic technology platform and cloud solution players. Additionally, the Accenture Innovation Architecture aligns us with pioneering start-ups and top academic institutions globally. We are constantly analyzing the state-of-the-art and readiness to identify potential breakthrough technologies for our clients.

A: Outside of government, where can public safety look for inspiration and best practices?

NS: Public safety professionals can be looking at the advances that we are making in self-driving cars. It's a robot in the form of a car, right? And it can detect so much of the environment around it. Law enforcement could look to that as a model for how AI can keep officers safe in complex situations.

They can also look at the legal industry, where AI is used very successfully for contracts management today, comparing a contract written with a certain law or rule book to say: “Okay, this contract is properly written.” In public safety you could leverage this to help keep up with changing rules and regulations.

A similar example is how the financial services industry is using AI to combat both fraud and cyber-attacks, both of which are needle in a haystack exercise.

Many public safety agencies have a significant citizen service mission. In this context, looking at the steps that consumer-focused enterprises have taken to integrate chatbots into the customer support function can be highly relevant. We are seeing sector leaders introduce automated customer experiences that are often preferable to traditional channels in terms of the personalization and speed that they can offer.

We should keep in mind that AI is not a black-box solution. You don’t just plug it in and—boom!—you are AI enabled. Any agency’s AI program must first have a clearly defined business context. Public safety will have to teach the AI to speak the language of public safety. That’s the next step going forward.

Nilanjan Sengupta

Associate Director – Technology Solution Architecture


Subscription Center
Stay in the Know with Our Newsletter Stay in the Know with Our Newsletter