The impact of AI/ML on cybersecurity functions

by | Jan 7, 2023

It is best to understand how machine intelligence continues to change the game for cybersecurity design, implementation, and operations with the ability to analyze massive security logs to expedite incident response operations. The alarming rate of sophisticated cyberattacks has triggered the adoption of AI-enabled programs in cybersecurity with the primary aim of ensuring privacy and data protection. However, the major concern for many is how AI will impact cybersecurity functions down the road and if AI will ever replace humans in cybersecurity.

Ray Kurzweil predicts that AI will reach human-level intelligence by 2029 and will surpass that of human intelligence by 2045, which is likely to result in human biological machine intelligence of our civilization a billion-fold [1]. Kurzweil claims we will reach singularity by that time frame. Is this the period for humanity 2.0? These predictions have serious implications for where the control of machines will drive various applications including cybersecurity.

AI and a paradigm shift
Is AI responsible for a paradigm shift from the era of humans in the loop (HITL) to humans on the loop (HOTEL)? How are security operations and optimization impacted by a situation where a machine, a server, or a computer system is capable or incapable of offering real solutions without the need for human intervention? With HOTL, machine-level intelligence does not necessarily need human intervention to offer solutions required to facilitate decision-making to solve real-world problems. In other words, the impact of automation driven by AI-based algorithms cannot be overemphasized. In the training of machine learning (ML)/ deep-learning (DL) models, both HITL and HOTL have practical significance with full-blown model accuracy and optimization. How do both concepts leverage both human and machine intelligence to create models? For example, with the HITL approach, humans are directly involved in feature engineering, model training, tuning, and testing with a particular ML algorithm using supervised learning or unsupervised learning. On the other hand, the concept of HOTL is perceived to be a good candidate for the training and testing of unsupervised DL/ML models. This paradigm shift will likely be controlled by the era of singularity and machine superintelligence.

As AI technologies such as ML, whether supervised or unsupervised and natural language processing continue to take the center stage in cybersecurity solutions, analysts will in tend to be empowered to swiftly respond to threats with a higher level of confidence and precision. As we dive into the future, the advent of AI will keep up with changing cyber defenses. However, the question of whether AI will eventually take over cybersecurity functions with little or no human involvement is yet to be known.


[1] R. Kurzweil, “Tracking the acceleration of intelligence,” 2021. [Online]. Available:

Would you like to share?

error: Alert: Content selection is disabled!!