Artificial Intelligence Solves the Common Problems of Security which is Inconvenient and Inefficient
By: Seongbin Choi, the head of Suprema R&D Center, Suprema Inc.
In ‘Knight Rider’, ‘KITT’ listens to people and finds a fast route, and ‘Airwolf’ detects and identifies all types of aircraft. These are the imaginary machines familiar to those who grew up in the late 80s. At the time, people used to think of KITT and Airwolf as products of the imagination that only exist in fiction. However, around 30 years later, voice recognition-based navigation systems and image-based object recognition products are being easily found all around us. Whether we realize it or not, over the past 30 years, the world has been transforming the imagination into reality.
Artificial Intelligence — The Driver of Change
Artificial intelligence is driving these changes. AI-based voice recognition and image recognition enable machines to understand and recognize things better than humans.
Artificial intelligence, which first appeared in the 1950s, was briefly revived in the 1980s, then again fell off the radar. In the late 2000s, AI re-emerged and became a driver of the IT industry. Artificial intelligence requires large-scale data for training, but the relevant data was not available until the 2000s. By the late 2000s, things began to change with the spread of the Internet, smartphones, and then cloud-based services.
The Internet and smartphones enabled data collection from myriad sources to the cloud, and the advent of the Internet of Things accelerated data collection. With this massive data collection, artificial intelligence has done what we could only dream about in the past. And the term “the 4th Industrial Revolution” describes the emergence of, and the transformation led by, an industrial platform consisting of IoT (smartphones), cloud computing, data, and AI.
The Dilemma of Access Control
Now, let’s see today’s access control industry. Security is supposed to be inconvenient and inefficient. This is something everybody agrees on. Greater security comes with a more significant number of procedures, which increases inconvenience and inefficiency to the individuals who are handling the tasks. The situation is the same in the access control system industry. Conventionally security enhancement approaches require a thousand people to follow a procedure to prevent a one-in-a-thousand problem. Balancing security with efficiency is an irreconcilable dilemma for all security managers.
And so, what kind of system do these security managers envision? Perhaps a system that offers enhanced security with easy operation and usability. We have discovered that AI can make such dreams of security managers into reality.
Increased Convenience, Efficiency, and Safety
We could effectively save time and money by monitoring the specific area with a higher probability of errors occurring instead of monitoring the whole procedure where the problem occurs one-in-a-thousand. Also, if we could predict possible problems by detecting signs of symptoms, then we can enhance security without sacrificing convenience and efficiency.
Such intelligence-driven operations are possible through AI training in large-scale data sets. Suppose we can collect user activity data and train AI with individual-specific behaviors and patterns; in that case, we can detect unusual undesired behaviors and patterns that are highly likely to cause problem. In addition, if a problem occurs, AI can learn behaviors and patterns to detect the action beforehand and prevent the problem from occurring next time. This is possible only when AI is deployed, and trained with highly accurate access, behavior, and location data in the cloud, collected continuously through the access control systems.
One thing that needs to be clarified here is whether accurate access and behavior data can be obtained and how. Currently, most access control systems determine physical access through authentication/tagging as evidence of presence or attendance. However, some people may decide not to enter after authentication, or others may circumvent authentication by following in someone who has tagged. This means it is impossible to obtain accurate access and movement data from existing access control systems.
However, real-time location systems (RTLS) can be a game-changer by enabling accurate access and movement information. It allows you to identify cases where authentication/tagging is abused (leaving after the first tag or following others without authentication). RTLS-enabled access control allows us to collect data with accuracy and quality sufficient for training AI.
Recently, smartphone manufacturers such as Samsung and Apple have been competing to make the best use of UWB, one of the typical RTLS technologies. If UWB is integrated with mobile credentials in smart devices, it can be easily deployed to access control systems. In the coming years, various RTLS-enabled access control systems with precise location intelligence will be available on the market.
In the future, we will realize what many security managers have dreamed about: access control systems that predict possible problems based on accurate entering in and out data preventing problem occurrence.
In the middle of the 4th Industrial Revolution
In 2018, I visited the Salesforce.com head office in San Francisco as part of the Korea Information and Communication Agency’s overseas training program. What caught my eye was a banner for Einstein from Salesforce.com service, covering the entire building starting from the entrance. Einstein is an AI-powered service that leverages the big data of Salesforce.com to recommend customers with a high possibility of conversion and forecast future sales volume to take preemptive measures. I was kind of skeptical about the feasibility of the technology. But it was simply enormous; AI trained massive amounts of data called big data outperforms humans, and cloud services powered by these AI engines deliver new value that on-premise solutions cannot provide.
Simply put, the industrial revolution represents a change in creating exchangeable value. That is a change in the way we make money. I think I saw how the way to make exchangeable value was changing in San Francisco in 2018, which helped me understand why such changes are collectively called “the 4th.” This also explains why developers in Suprema, including myself, are dedicated to AI to improve the convenience and efficiency of cloud-based access control offerings.
