Automation gone wrong—and right

September 7, 20183 minute read

Select article text below to share directly to Twitter!

Dismiss

In the wise words of Captain Jack Sparrow, there are only two rules that matter: “What a man can do, and what a man can’t do.” But is this true for technology?

In a world where autonomous cars can’t read stop signs and a new generation of automated personal assistants imitate John Legend, it’s high time to pose the question: Are the rules for automation and artificial intelligence (AI) about what it can do—or what it should do?

Understand that artificial intelligence is artificial

In truth, the answer is a little of both. AI definitely deserves a seat at the table, but it has a history of failing when it comes to ethics and judgment calls. Despite the buzzword’s popularity, artificial intelligence is still just that: artificial. AI programs can only extrapolate from what they already know, which means if the data set they’re trained on is inadequate for the task at hand or the algorithms they’re initially programmed with are inaccurate, AI is next to useless.

For example, it took the American Civil Liberties Union of Idaho more than five years and more than $50,000 to figure out why a new algorithm was reducing people’s Medicaid assistance by 20 or 30 percent. It turned out the algorithm and data set were so inaccurate a judge ruled them to be unconstitutional.

It isn’t just in matters of ethics where AI fails; relying on AI for your security needs isn’t always the best idea, either. Since it can only extrapolate, AI can’t always detect attacks on unknown vulnerabilities—also known as zero-day vulnerabilities—and, therefore, can’t always find the advanced threats AI security tools are built to look for. Besides that, most cyber attacks don’t actually target zero-day vulnerabilities. According to Gartner, 99 percent of vulnerabilities exploited by 2020 will still be known vulnerabilities. An AI tool designed to search for unknown threats may not be what your organization needs, after all.

If AI isn’t the best way to take the stress out of cybersecurity, then what is? That may be where automation comes in.

What’s faster than a speeding bullet? Automated security

It’s easy enough to say, “Patch your systems!” but as many IT pros know, it’s rarely that simple. Many companies struggle to map their existing technological landscape, let alone their vulnerabilities. Fortunately, that’s where automation can make all the difference.

Where it would take a person countless hours to track down and inspect all the endpoints on a network—especially in today’s world of BYOD—automated network monitoring solutions can identify vulnerabilities before they’ve been exploited. Better yet, imagine if your endpoints could inspect themselves—and recover after discovering a problem. Instead of checking every laptop, mobile phone, or business copy machine yourself, you’d be able to rest assured they, at least, were secured. Some innovative printers are making that a reality with built-in self-healing technology—an example of automation at its finest. In addition, modern printers can seamlessly integrate with many security information and event management (SIEM) tools, so you have the option to keep the human in the loop.

Form the perfect team with automation

You never want to rely on automation alone for your IT security; it’s not a magic bullet. But it’s a great tool for executing instructions quickly and making sure known vulnerabilities aren’t exploited. It can also prove to be a more reliable alternative to AI in a world where AI is still getting its footing.

It’s only when you combine the efforts of humans and machines that you can really get the most out of new security technologies. Automated security technologies, whether they’re integrated with your network or built into your business copy machine, make that ideal a reality.

  • Recommended for you
  • Recommended for You