When Will AI Become a Reality in Security?
For all the talk in recent years within the video surveillance market about how artificial intelligence (AI) and machine learning is going revolutionize the industry, even the most ardent supporters of the technology would admit it’s not ready for widespread adoption. The vast majority of AI and machine learning solutions available today leverage deep learning algorithms that require massive amounts of data to be able to “learn” and classify objects with any degree of accuracy. But even with quality data sets, these next-gen video analytics are still prone to error and it will require more time and trial and error for vendors to further refine their offerings.
So, with AI not yet ready for prime time, the question for systems integrators is: how do you balance selling and installing today’s solutions while staying prepared for the next generation of analytic technology when it becomes ready for mainstream adoption? In a keynote address at the 2019 Milestone Integration Platform Symposium (MIPS), Tim Palmquist, VP for the Americas at Milestone Systems, said this dilemma was succinctly summarized in a comment he received regarding his keynote at last year’s event.
“The person said all of this talk about machine learning and AI is great but what can I sell today? My first instinct in seeing that comment is that this person doesn’t understand why we come together at MIPS. After all, we’re looking into the future and while we recognize what’s going on today, well that’s not the focus, right?” Palmquist says. “The comment has stuck with me and I’ve come to conclude that that comment is not an outlier but it’s very appropriate and practical. All of us together have businesses to run, revenue targets to hit, commissions to earn, profits to make, and we do all of that with the technology and products available today.”
In addition to viewing the market through the lens of what can be sold today, Keven Marier, Director of Technology Business Development at Milestone, said he encourages the company’s partners to approach selling and installing security systems in terms of what kind of value can they create for customers tomorrow. In a separate keynote at MIPS 2019, Marier discussed the curve of various technologies in the security industry and where the adoption of emerging and contemporary solutions is currently.
Marier described the process that many integrators find themselves in today as “orchestrating” intelligent devices and systems as part of a journey with a destination that ultimately lies in the manifestation of what he referred to as the four “A’s:”
- Autonomous Devices (drones, smart robots, machine vision and learning, AI governance, etc.)
- Aggregated Systems (compute infrastructure, deep neural networks/learning, AI Platforms-as-a-Service)
- Automated Solutions (predictive and prescriptive analytics)
- Augmented Operations (augmented and virtual reality, virtual assistants, conversational user interfaces/bots)
Of course, no one can say exactly when these technologies will become commonplace in organizations as they are essentially in their infancy today. And, as evidenced by previously over-hyped technologies, timing is of particular importance in the security industry.
“Is it one year or is it 25 years away? We have a long history of companies that started too early – ObjectVideo, IPIX which invented 360-degree cameras and in 1999 went public for $73 million only to go bankrupt in 2006,” Marier says. “Timing is so important in business decisions but so is execution.”
In looking at the Gartner Hype Cycle for AI, Marier says it’s clear that the technology is in the embryonic phase with many solutions not expected to mature or plateau until five to ten years from now and even more than 10 years in some cases. Other technologies, however, such as robotic process automation, are expected to mature within the span of less than two years.
Though the maturity of some of these technologies is still a ways off, Marier says that integrators should be having conversations with their end customers now to determine how and where they might be applicable in the future. “When we want to start talking about creating value for our customers, we need to say, ‘where do you want to go?” You need a map and you need some directions,” he adds.
By way of comparison, when you look at the hype cycle for physical security, Marier says some technologies that have long been adopted as mainstream solutions in the industry, such as video management software and NVRs, have only recently entered the curve’s “slope of enlightenment” whereas services like VSaaS (Video Surveillance-as-a-Service) are entering a period that Gartner dubs as the “peak of inflated expectations.”
So, while some may believe that widespread adoption of AI within video surveillance is a pipe dream, the reality is that industry has only begun to scratch the surface of what is possible. Even migration to the cloud, which Marier says is the most important conversation that integrators can have with end-users today, is still is in its early stages in physical security.
“There’s a rule in this that I call ‘plus 10’ and that means when the innovation cycle happens, it is at least 10 years before we start to see mainstream adoption,” Marier adds. “So, what can you sell today; orchestration and planning. How do I create value for tomorrow; moving your customers’ businesses from one innovation cycle to the next.”
About the Author:
By Joel Griffin, Editor of SecurityInfoWatch.com and a veteran security journalist. This article appeared first here.