Dr Daniël Reichman, CEO and Chief Scientist of Ai-RGUS examines the advantages and limitations of enhancing system integration with AI.

Whether it’s running a cybersecurity program on your business computers or testing a video surveillance security system, security-based AI is becoming increasingly used in everyday life. Security AI software is also transforming the way security is involved in businesses and organizations and updating system integration.

With advancements in AI algorithms, there is a growing trend towards incorporating AI into security systems to improve effectiveness and efficiency. Unlike traditional software, which relies on human programmers to write instructions, AI software learns its own instructions from labeled data examples. In this way, it can identify complex patterns and concepts with accuracy and with much less human involvement.

While AI holds great promise in enhancing system integration, there are still some limitations to its implementation. As with all technology, AI algorithms are improving every day, leading to tools and software that can more seamlessly integrate into the security ecosystem, helping system integrators better service end users. System integrators who understand the end-user’s scenario, regulations and needs can create a customized plan to not only ensure the effectiveness of the system, but also help to prove its value over time.

Here is how AI is helping system integration better meet the needs of end users.

Traditional software versus AI software

Traditional software – built without AI – relies on a computer programmer to write rules or instructions to compute a certain quantity, such as adjustments to payroll based on time off, commission, bonuses or preparing a certain deliverable, like filling out a form or creating a PDF report with some data.

AI software is characterized by a rule engine that was not entirely derived from a person. This means that in some way (and in most cases), a dataset was used to identify the set of rules that most highly correlated with a certain concept.

The reason this is meaningful is because AI is writing out the rules. Imagine trying to write out rules to describe a natural object like an animal or type of landscape like a hill. The ability of AI to write out rules relating to the target concept, with such accuracy that it rarely confuses other concepts with the target concept, is the power of its intelligence. This ability opens the door to identifying objects, concepts, behaviors and patterns of a wide range of complexity.

However, AI isn’t yet applicable in every use case. Today, I would not use AI software to write out a payroll program because of how precise the rules need to be. Likewise, I would never task a person to write out a program by hand that recognizes “lions” in images because of the overwhelming number of nested rules that would need to be written.

The main reason AI is crossing into security systems at a rate we are seeing today is in its ability to support the goal of maintaining safety and security. As pattern recognition improves the capabilities of AI, tasks will begin to take less time and new potentials will start to emerge that would have been impractical prior to these new capabilities.

Remaining limitations

Incorporating AI into a security system adds a layer of “recognition in the data” beyond the data that traditional software makes available in graphs and online dashboards. Examples include facial recognition for door access events, making it easier to identify whether an access card is being used by the person to whom the credential was provided, which may altogether eliminate the need for physical credentials. Further examples include automatically identifying potentially malicious patterns such as loitering, entering a prohibited area or moving an object from a predetermined location and automatically identifying if a camera was titled or if its view is blocked.

With the preponderance of available data, AI systems can already be quite accurate, meaning that AI’s knowledge of the concept in question can already be precise without additional work. New developments in tuning AI algorithms have also improved to the point where additional on-site tuning will focus on major readjustments, rather than minor, tediously repetitive changes. A lower setup time means greater margins and fewer training requirements for those integrators.

That said, there are some security professionals who are still hesitant about the wide implementation of AI to improve their work processes. There has been a breakdown in communication between AI vendors, security professionals and end-users over the years that has definitely caused friction. AI is statistical software, which means there is a margin of error around any statement of accuracy or fitness of purpose. Furthermore, it is neither practical nor possible to collect data for every scenario for testing purposes. Therefore, working with reputable professionals and AI manufacturers is critical.

The opportunities of AI software for system integration

As AI technology improves, new opportunities for improved service emerge. One such natural opportunity lies in the maintenance of the system. System integrators are often the point of contact regarding system maintenance.

Offering system maintenance can be costly for a variety of reasons. Firstly, it could require sending a person on-site to resolve the issue which can incur variable costs. Secondly is the difficulty in identifying that there is a problem at all. Camera systems are complex and have many parts, while having to work all day every day, across the seasons.

Finding out that a camera has not been recording for a while or that its view is obscured or has been tilted is not an option, however. With emerging AI technology, it is now possible to receive alerts for these conditions automatically. This eliminates a gap in coverage which may lead to hazardous situations. Including remote health monitoring software when deploying a camera system leads to a more responsible design and allows the end-user to get the full value out of their system.

Trusted system integrators are those who take the time to understand the end-user’s scenario, needs and regulations to suggest a customized plan. This approach not only ensures the effectiveness of the system, but also helps maintain the system’s longevity. By proactively identifying and fixing any issues, system integration can minimize the likelihood of unexpected breakdowns and minimize downtime, resulting in a more secure and reliable system that meets the needs of the end-user.

A software solution that facilitates this process allows system integrators to be more efficient and cost-effective in their services, providing added value to their clients. The importance of end users to partner with a trusted and experienced system integrator, who not only provides superior equipment but also prioritizes the maintenance and upkeep of the system, should be a given.

The road ahead for system integration

There are some technological hurdles ahead for improving system integration. For example, updating legacy systems is often a manual process. Even new software is not always built with automation in mind, causing added work for the system integrator. Despite these challenges, the benefits of having a well-integrated and well-maintained system are undeniable.

By overcoming these technological hurdles and investing in the right tools and resources, system integration can improve efficiency, enhance overall performance and provide even greater value to the end user. In today’s fast-paced and ever-evolving technological landscape, staying ahead of the curve is crucial and system integrators play an essential role in ensuring their clients are keeping up at the same pace. By embracing innovation and continually improving processes and techniques, system integrators can continue to provide state-of-the-art services and deliver the highest value to their clients.

Dr Daniël Reichman is the CEO and Chief Scientist of Ai-RGUS, an AI startup spun out of Duke University. After receiving his PhD at age 25, Daniël founded Ai-RGUS. With over 24 university publications, he obtained his doctorate in Electrical and Computer Engineering from Duke University from a program fully funded by the US Army Research Office. He also successfully completed the first two actuarial exams and obtained a minor in Mathematics.

This article was originally published in the April edition of Security Journal Americas. To read your FREE digital edition, click here.