Real words or buzzwords?: Operationalizing AI

July 18, 2023
AI is not a product, but a broad category of software that enables products and systems to do more than ever before possible. How do we put it to good use?

Editor’s note: This is the 70th article in the “Real Words or Buzzwords?” series about how real words become empty words and stifle technology progress.

"Operationalizing" AI in security operations involves integrating artificial intelligence into the routine operations of a security program. While AI has been heralded as a "game changer" for the physical security industry, it's crucial to recognize that it's not a panacea for all security operation deficiencies.

Security operations encompass human, procedural and technological components, all of which must work together to form an efficient and effective system for physical security.

Part of the power of today’s physical security technology is that a security system is actually a system of systems, as many security devices are intelligent systems unto themselves – such as intelligent network cameras.

Breaking Historic Limitations of Security Systems

Historically, security operations relied heavily on physical barriers, security officer observation and inspection stations and patrols. The advent of video cameras in the 1960s and card access systems in the 1970s marked a shift towards technological solutions, reducing the need for guards.

However, for decades, the balance between cost and security efficacy has been compromised by technology constraints, high costs and human limitations. What sets the emerging era of security capabilities apart are five parallel technological trends:

  • Exponentially increasing technological capabilities
  • Exponentially decreasing technological costs
  • Ubiquitous computing (it’s everywhere)
  • Increasingly data-rich cyber-physical systems
  • Increasingly capable automation of systems and devices at any scale – automation that learns and is capable of autonomous operation  

These trends, just beginning to transition from their infancy, have broken past the historic limitations of security technology deployments and are revolutionizing the future of physical security systems. (See Real Words or Buzzwords?: Physical Security Watershed Moment.)

AI is the pivotal element propelling these advances. Technology costs and capabilities no longer pose a barrier to security system effectiveness. The primary limiting factor is our ability to effectively integrate AI into existing deployments of people, process and technology.

We are hyper-focused on the technology bits and pieces, when we need to be focused on the complete picture of people, process and technology to effectively operationalize AI.

Operationalizing AI

The application of AI in physical security involves three key technology areas, each of which impacts security operations in different ways:

  • Reliable Analytics: These comprise audio, video and sensor-based analytics leveraging machine learning and deep learning to automate tasks traditionally requiring human intervention.
  • Intelligent Devices: These are security devices equipped with machine learning and deep-learning capabilities. They can run analytic software and provide situation metadata to overarching security operations platforms.
  • Advanced Situational Awareness Platforms: Data integration from multiple devices and system sources is enabling early risk detection. Real-time data fusion is enabling a shift from reactive to proactive security operations.

The security operations capabilities that are now being made possible by these advancing technologies have never existed before.

I asked James Connor, Head of Corporate Engagement for Ambient AI – one of the companies leading the transformational use of AI for physical security system technologies – “How do we start taking immediate advantage of what AI-enabled technologies have to offer?”

AI is transforming security and safety monitoring and surveillance with computer vision that can automate tasks that currently require countless hours of human analysis, as well as provide more actionable data in real time.

Connor provided six experience-based practices for deploying computer vision on the front lines of security operations.

1) Start Small, But Strategic

Look for operations where AI can alleviate repetitive and tedious human review of video feeds and integrated alarms that require video verification, or others that generate false positives that can be adjudicated with the application of video AI. Analytics on high-volume security camera footage is a prime target. Begin with well-defined specific use cases versus trying to mimic all human visual abilities.(Small does not mean limiting the number of devices. It means choosing a focused target for improvement.)

2) Supplement Staff, Don’t Supplant

Position AI as enhancing existing personnel by offloading tedious work such as alarm acknowledgement or clearing and enabling staff to focus on higher-level analysis. Make sure staff are on board and understand the AI augmentation approach.

3) Verify Results

Have staff validate a sample of AI-generated analysis for quality assurance and feedback. Understanding AI model performance based on human review will improve accuracy over time and ensure that all expectations for performance are aligned.

4) Monitor Closely

Watch for privacy violations, false positives, unexpected failures and adversarial attacks attempting to dupe the system. Anomalies in detection on the AI output can flag unusual results for human review and address issues that are a result of the environment or infrastructure such as lighting or latency.

5) Keep Data Flowing

Computer vision models improve over time and vast, varied training data representative of the deployment environment will continue to enhance their abilities. Maintaining an ever-increasing data pipeline from operations into model for refinement through operational reviews will be key to early success. Data diversity is key to reliability, and this is where operations personnel can drive performance.

6) Maintain Human Oversight

While AI can do amazing things, human judgment is still essential for complex situations with ambiguity. Keep experienced staff in the loop for contextual decision making and model governance.

The right integration approach can make computer vision AI a force multiplier for security. When thoughtfully implemented, it will propel operations into a new era of efficiency and effectiveness.

Special Note: The Global Security Operations summit being held at LinkedIn Global Headquarters Aug. 16-17 in Sunnyvale, Calif., is dedicated to helping security practitioners transition from legacy security operations thinking to a future-ready security mindset. You’ll be able to see and hear from several AI experts and talk to customers who have effectively deployed AI-enabled products and systems. End-user customers with large-scale physical security system deployments should prioritize attending this summit.


Ray Bernard, PSP CHS-III, is the principal consultant for Ray Bernard Consulting Services (RBCS), a firm that provides security consulting services for public and private facilities ( In 2018 IFSEC Global listed Ray as #12 in the world’s Top 30 Security Thought Leaders. He is the author of the Elsevier book Security Technology Convergence Insights available on Amazon. Follow Ray on Twitter: @RayBernardRBCS.

About the Author

Ray Bernard, PSP, CHS-III

Ray Bernard, PSP CHS-III, is the principal consultant for Ray Bernard Consulting Services (, a firm that provides security consulting services for public and private facilities. He has been a frequent contributor to Security Business, SecurityInfoWatch and STE magazine for decades. He is the author of the Elsevier book Security Technology Convergence Insights, available on Amazon. Mr. Bernard is an active member of the ASIS member councils for Physical Security and IT Security, and is a member of the Subject Matter Expert Faculty of the Security Executive Council (

Follow him on LinkedIn:

Follow him on Twitter: @RayBernardRBCS.