Ever since computers entered the mainstream, there has been anxiety about machines becoming capable of intelligent thought. Early artificial intelligence (AI) programs, which could solve logic problems and speak phrases, spurred optimistic predictions of AI takeovers by the end of the 1900s. Obviously, those projections didn’t pan out, but there is still concern that AI will eventually become sophisticated enough to threaten mankind.
In many respects, AI is still far away from reaching such potential. For example, you couldn’t place an AI-powered robot in a random home and expect it to successfully make a cup of coffee. However, there are other domains in which the gap between man and machine is narrower, and the two can almost be regarded as peers — namely, in IT security solutions.
Why an AI-human combination is needed in cybersecurity
Effective defense requires the ability to accurately identify a wide range of threats. Human experts and AI programs each excel at particular tasks, such as:
- Automatic screening based on security policies: AI-based systems are the best at combing through massive data sources, such as logs of network events, and detecting anomalies to block known threats. Most data in existence was created this decade, and the only way to keep up with its pace of generation is through automation.
- Analysis of odd and unusual behavior: Humans have the advantage here. For example, a long-time employee downloading a bunch of files before a conference call might get flagged by an automated defense system, but a human would likely better know if he or she is trustworthy.
- Sorting false positives and negatives: AI-driven security produces a lot of questionable results, partly because it can’t always distinguish between normal and suspicious behavior. False positives are costly, with the Ponemon Institute estimating the typical organization loses $1.3 million to them annually.
Relying exclusively on either humans or AI programs is not feasible, at least if comprehensive cybersecurity is the goal. The ideal combination would instead pair experienced personnel as well as trusted external partners with up-to-date, AI-enhanced security infrastructure. IT security solutions adhering to this model are sometimes categorized as “hybrid AI” platforms.
The benefits of hybrid AI in security contexts
A few years ago, MIT researchers tested a system combining machine learning and human insight. It predicted cyberattacks with an 85 percent success rate, or three times better than earlier alternative systems the team had tested, while reducing false positives fivefold.
Combining man and machine for cybersecurity purposes addresses the shortcomings of each side. Human-centric security systems often cannot scale to account for all possible attack vectors; AI-dependent ones in contrast generate many false positives.
Although the benefits of a hybrid implementation are apparent, the question is how to set it up. A trusted partner such as LaSalle Solutions can guide you through a security assessment, installation, configuration and ongoing review process, so you get a reliable solution closely matching your requirements.
Learn more by visiting our Security page today.