Should AI Be Allowed to Make Independent Security Decisions?

[johny899]

AI can be both rewarding and terrifying. I can see AI stop cyber attacks immediately and think to myself "this is incredible". Then, I ask myself "should AI be allowed to make independent security decisions?" Tough question, right?

Why People Like the Concept​

People prefer AI because of the speed at which AI operates compared to a human being. We need to eat, sleep and take breaks. AI is always "on" like an assembly line.

This is why AI appears to provide a lot of value in Security:

• AI operates in real time to identify threats
• It has the capacity to analyze vast quantities of data quickly
• AI learns and gets better daily.

How often have you waited on an alert for your security only to think "that took too long"?

AI has eliminated that pain point. Speed is critical when it comes to preventing attacks.

Here Is The Dilemma​

Artificial intelligence (AI) constitutes a major challenge because machines may misinterpret human behavior as a threat. Eventually, it can automatically disable access to that machine's resources and functions.

Consider a scenario on payday when no employee can access their payroll due to the AI error.

Possible errors made by AI include:

• Blocking legitimate users due to its failure to understand legitimate behavior
• Allowing malicious users to mislead the AI
• Enabling the AI to perform actions that the user cannot reverse.

The potential of this type of misidentification makes it extremely troubling.

My View​

I believe AI can support humans with their security decisions. However, I do not believe it should act independently. Instead, AI should be a "smart assistant" in making informed and responsible security decisions.

The intelligent assistant must guide the user, by identifying the potential threat and recommending action. The human will accept or reject the advised action based on their level of knowledge and experience.

Thus, I find this safety promise appealing and logical.

What To Conclude​

Based upon the preceding discussion, I do not endorse allowing AIs the ability to independently make security decisions. Rather, I support AI's role as a supplement to human security decision-making. However, humans will ultimately be responsible for all actions taken based on the advice of an artificial intelligence.