Innovation is constant, and it drives continuous improvements in technology. This is based on a number of factors, primarily the amount of money spent on research and development, as well as Moore’s Law, which states that processors become faster every year. As technologies mature and surpass the capabilities of previous-generations, we see sudden shifts from spending on an older technology to spending on the latest technology.
A prime example of this phenomenon is the shift from analog to IP-based video surveillance – which took many years to develop and mature; however, when the shift finally occurred, it was relatively quick to the extent that few of the incumbent players were able to keep up.
When it comes to innovation, the biggest challenge is not being able to recognize that a change is coming – with just a little knowledge, practically anyone can speculate and make a prediction about future trends with relative accuracy. The biggest challenge with forecasting is actually to get the timing correct. Is a certain technology going to change the market this year? Next year? In 10 or 20 years?
If a company invests and positions itself correctly – and correctly predicts the timing – the benefits are vast. If it is too soon, the company will burn through all their funds before sales takes off, and the initiative (if not the company) may be shut down. If the company is too late, it may become stuck in a commodity race in which margins are razor-thin at best.
What are the trends to watch in the security industry today? Many of them are megatrends from other industries, such as the cloud, big data and artificial intelligence (AI). The latter of these is tremendously interesting to study more in depth, as it has had its hype cycles several times and is going through another one as we speak. Has the time come for AI in security, and, what is the best way to play into this hype?
In Depth: The AI Hype Cycle
Let’s get something out of the way: AI is not a new trend, and it is certainly not specific to the security industry. All companies are pouring money into making better use of ever-increasing amounts of data and intelligence provided by better computer chips and more sophisticated algorithms.
When I went to university, I had friends who were researchers in AI and who talked about it as the next big thing. That was back in 1988. I also had a colleague send me an Economist article recently, which wrote excitingly about AI and its impact as the next big thing…the article had been published in October 1995.
It is hard to avoid reading about self-driving cars, and while they are definitely equipped with great self-driving capabilities – powered by AI – as compared to the cars of yesteryear, the question remains: When will be see a society where most cars are self-driving (which would be lovely for all of us with long commutes)? Next year? In five years? In 10 years? In our lifetime? It is hard to say for sure.
The “old technology” that AI has set out to replace is something we all have: some kind of function performed by the human brain. The question is when technology is capable to fully replace a certain task or function – and that depends on the accuracy required.
Case in point: With the impact of social media, Facebook is working diligently to clean up fake news, hatred, etc., and the company needs to sort through and examine all text that is posted every single day. While text is quite easy to parse through – and you would think that there should be some automated AI algorithms to do it – the solution Facebook took was to employ thousands of people to parse text manually to ensure the appropriate accuracy. Thus, there are still many relatively simple tasks AI cannot fully perform, even within a technologically sophisticated company.
What does this all mean for the security industry? For starters, we are undoubtedly in the midst of another analytics and AI hype cycle, seizing on the general AI megatrend. The aim of AI in security is to augment human operators to increase safety and efficiencies, which makes sense, as more than 90 percent of security resources are still spent on humans, according to ASIS research.
The last time we had this hype was about 10 years ago, when hundreds of millions of dollars went into funding new analytics startups – few of which remain in existence today. As referenced, there was also a general AI hype in the mid-1990s, so it seems to come back around every 10 years.
Naturally, there is a high level of excitement this time around, and it would be tremendous if AI actually works out. And like with every other hype, nobody wants to be left standing on the sidelines or be a naysayer. This is similar to the way hype is greeted in the stock market – people buy into the hype cycle because they do not want to miss out on a potentially lucrative opportunity while others are taking advantage of it. Sometimes it works out and sometimes it does not, which is true of any hype.
The Difference in 2018
Unlike what the security industry – the video surveillance market in particular – saw with the last big video intelligence hype about 10 years ago, this time there has not been a slew of start-ups undertaking the AI applications. Rather, it is the big vendors who are investing money or have bought up small start-ups.
One positive difference from that last hype cycle is that these large vendors who are getting into the AI game will most likely still be around in five or 10 years; however, the downside is that it is more difficult to accurately track how the AI market is really performing. If you were to ask a vendor how many installations involve deep learning, the answer might be “most installations” – even though in reality there are very few cameras in each installation that are using intelligence. This only serves to further muddy the waters and make it more difficult to gauge the potential opportunity that exists with AI.
All this aside, AI is definitely improving and can be incredibly useful in some applications, depending on requirements and expectations. For example, license plate recognition has proven valuable in several situations such as parking garages and ports where the technology can be used to provide entry and exit from particular areas while maintaining a log of arrivals and departures. Similarly, retail analytics deliver tremendous value by providing insight into basic trends by counting people and heat-mapping. Facial recognition in controlled environments, such as passport control, also works pretty well.
As technologies continue to improve, it is likely that once again, this big AI hype will slowly fade in security – and other industries – as reality sets in. But rest assured that there will be some other cool new hype – or hypes – to take its place as excitement continues to build for the future of security.
Of course, we would all love for these hypes to arrive and mature faster than reality would have it; therefore, it is essential to continue to evaluate each hype cycle and the opportunities it presents in an effort to seize on the “next big thing” for improving security before it can pass you by.
Fredrik Nilsson is VP, Americas, for Axis Communications and is the author of “Intelligent Network Video: Understanding Modern Video Surveillance Systems” published by CRC Press and now available in its second edition. Request more info about Axis at www.securityinfowatch.com/10212966.
