forbes.com | June 1, 2018
Artificial Intelligence (AI) and Machine Learning (ML) are extremely hot topics these days — both in the real world and in Hollywood.
In computing, the terms artificial intelligence and machine learning pop up frequently in any discussion of big data and analytics, with some people erroneously using the terms as if they were interchangeable.
To use an animal analogy, AI and ML look alike as much as a horse resembles a zebra, but that’s where the similarity ends. Artificial intelligence refers to the ability of machines to perform tasks that require intelligence. Machine learning is a subset of AI based on the concept that technology can enable computers to learn and adapt through experience. ML emulates human cognition by basing learning on experience and patterns.
Some observers predict a dire future, where the dark side of artificial intelligence will create a paradigm shift and the balance of control will move from humans to machines. Their darkest vision is “total eradication of the human race” via the Singularity. Pragmatists (or optimists, if you prefer) believe such a scenario is hype, while acknowledging that AI will play an increasingly vital role in our lives.
AI Is Everywhere
The rapid evolution of technology has influenced our quality of life in ways we never anticipated. The impact continues to grow, fed by big data, analytics and the application of AI to software.
Artificial intelligence, for example, is at the heart of Apple’s new iPhone. The iPhone X has a neural engine built into the processor. The engine accelerates specific types of AI software (i.e., neural networks) that were built to process pictures and speech.
In his book World Without Mind: The Existential Threat of Big Tech, author Franklin Foer observes that “We are merging with machines in an unprecedented way.” Automation, he points out, compels us to forgo thinking of performing certain manual actions (like autocorrect in typing) or engaging particular mental functions (like online shopping choices).
“Let the machine do it” has become an ingrained attitude for many of us. Google Maps uses AI to show people the shortest route to a destination and how to drive it in the shortest time possible.
Amazon and other internet-based enterprises rely on artificial intelligence to post online ads and send people marketing emails based on people’s browsing history. Such enterprises use data analytics of previous purchases and online product research to customize marketing material with the goal of influencing people’s next purchases.
Machine Learning Is Approaching Human-Level Intelligence
Machine learning is probably the best way to progress toward human-level intelligence, as it uses an iterative, automated approach that enables it to analyze data until a clear pattern is found. This allows it to go beyond looking for “known” or “common” patterns to make sense of the unknown.
By applying machine learning to behavioral analytics, organizations can make sense of the large volumes of data generated by infrastructure platforms, application and security products. In this way, ML can be simultaneously applied to hundreds of thousands of discrete events from various datasets so meaning can be determined from risky behavior patterns — and used as an early warning prevention system.
Machine learning can perform link analysis to evaluate relationships or connections between data nodes or objects. The technology identifies relationships among various types of data nodes such as employees, access, transactions and security alerts.
While many people talk about artificial intelligence as having arrived in the security technology landscape, this isn’t the case with security analytics. Mature ML is certainly making strides. Analytics solutions driven by machine learning are getting smarter the more they are refined and used. But they still do not think for themselves — not yet at least.
The primary obstacle for artificial intelligence is that it still lacks human contextual judgment to adapt to changes in the constantly evolving security landscape. Understanding what works today may not work tomorrow — or what did not work today may work tomorrow — is a subtle difference that humans can interpret and machines cannot. Training AI to make this leap will be a big step forward for security.