Fredrik Nilsson is Vice President of the Americas for Axis Communications, overseeing the company’s operations in North and South America.
Everyone wants in on the AI gold rush, and security providers are no exception—but potential partners and customers don’t always have a thorough grasp of the technology or its real-world applications.
As promising as AI may be, how is it being used in practice? What are the risks, and how are they being addressed? And, perhaps most importantly, how are vendors addressing the ethical and privacy concerns that come along with AI? Below are the five biggest questions security vendors are hearing—and how to answer them.
1. What are the public safety applications for AI?
The security industry has been using AI for years—they’ve just been calling it “analytics.” With the volume of cameras in use today, manually monitoring them is an impossible task, which means organizations are increasingly turning to automated solutions.
Roughly two years ago, deploying deep learning capabilities at the network edge became possible, effectively revolutionizing video analytics. Organizations no longer had to send raw video to the cloud or store massive amounts of footage, both of which could be prohibitively expensive. Instead, AI-based analytics could automatically detect and alert to incidents while generating metadata for further analysis.
Modern analytics can detect suspicious or dangerous activity with significantly greater speed and reliability than humans. By reducing the role of manual processes in security, newer systems with deep learning-enabled cameras are helping businesses improve the accuracy of their incident detection processes while also ensuring security budgets are being deployed effectively.
2. Is AI only useful in certain industries?
Most companies can benefit from AI in some way. The growing accessibility of video analytics brings benefits that extend beyond security to provide insights into operational efficiency and business intelligence.
For many businesses, analyzing sales data, customer traffic, quality in production and other information can lead to insights that positively impact the bottom line. The rise of generative AI solutions has even made it easier for businesses to streamline repetitive tasks like drafting emails, responding to customer support queries and more.
Still, it’s important to recognize the limitations of technology. At the start of the Covid-19 pandemic, there was a rush to implement thermal cameras to enable fever detection for customers visiting physical locations, but the technology simply wasn’t there.
As useful as thermal cameras can be, fever detection analytics were not accurate enough to provide the desired value. As a result, people lost faith in the technology because they approached it with false expectations. When it comes to AI, it’s critical to understand what you want to accomplish—and whether it is actually possible.
3. How can businesses use AI while complying with privacy regulations?
Regulations are steadily emerging that govern how AI can be developed and used—and chief among them are privacy regulations. But regulation always lags behind innovation, and it can be difficult to anticipate how different compliance frameworks will approach privacy or security issues.
As a result, it’s a good idea to be proactive rather than reactive, building privacy features into products from the beginning rather than waiting for regulators to provide instructions.
For video surveillance, that means implementing privacy masks to protect the identities of those being recorded, as well as cybersecurity measures that strictly monitor who has access to recorded video—as well as the devices themselves. This will ensure the company is in a good position to comply with future privacy regulations, whatever they may be.
4. How can AI be used in an ethical and responsible manner?
Training AI requires a significant amount of data, and where that data comes from is important. The debate over whether it is ethical (or even legal) to train AI on copyrighted material rages on, and inherent bias remains a significant problem in training data.
Organizations can purchase data sets to train AI models, but it isn’t always easy to tell what biases that data might come with (or if it complies with regulatory guidelines). If certain video analytics struggle to differentiate between individuals of specific ethnicities or genders, that’s a problem that can lead to serious ethical concerns.
These are questions that need to be answered before incorporating data into an AI training model. You can’t un-bake the AI cake: Once an AI solution has been trained on tainted data, it cannot be “untrained.” Organizations should use their own carefully vetted data sets to train AI models rather than turning to third-party data. Using data you control is the only reliable way to address ethical concerns.
5. Seriously, though: Is AI going to take over the world?
Maybe! But probably not. I recently read a fascinating book titled Sapiens: A Brief History of Humankind. Author Yuval Noah Harari points out that Neanderthals were not only stronger than Homo Sapiens but had bigger brains as well. How, then, did humans win out?
Harari posits that despite their bigger brains, Neanderthals lacked the ability to believe in abstract concepts. Humans can bond with one another over faith, beliefs, politics—even brand loyalties. The more individualized Neanderthals were left behind as humans built cooperative societies.
Why is this relevant? Well, it’s true that AI may someday be smarter than humans. In some ways, it already is. But humans have shown that we don’t have to be the biggest or the smartest to retain our place at the top of the food chain. As long as we retain our curiosity, our inventiveness and our ability to collaborate, AI probably isn’t going to take over the world.
Asking—And Answering—The Right Questions
As AI-based solutions are used for a growing range of security applications, questions about privacy and ethical considerations are inevitable. But when used appropriately and responsibly, AI can provide a wide range of benefits, both improving security capabilities and offering operational and intelligence advantages.
Asking the right questions is important—and so is implementing the right solutions. By taking a thoughtful and considered approach to implementing AI-based solutions, businesses can enjoy the technology’s many benefits without much fear of its drawbacks.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?