The Journey to Cybyr-AI:

This is about the challenging journey into the unknown that we are all facing. If generative artificial intelligence (GenAI) is not the topic of the moment in your business life, then you are living on another planet! 

Having written my first data-driven guidance and measurement software many years ago, it became obvious that if I wanted to create an AI-assisted implementation for my cybersecurity software, then I needed to understand the nature of the beast. I rapidly discovered that using GenAI alone was never going to cut it.

See cybyr.com/ainetworking  for development based on the Oct. 2024 article published in ISE magazine: “AI Networking: Training the Unruly Child.”

Why Is This Important, and What Inspired Me?

My goal was to create a cybersecurity system reflecting my organization’s purpose – Cybyr.com: Every Organization Protected. The system should make recommendations and measure risk reduction over time.

Much more than that, I wanted users from all types of organizations to be appraised of the very latest developments and ask for information GenAI style but not with the dated, risky large language model (LLM) approach and uncertain privacy.

Therefore, I embarked on a different path that enabled the use of AI while recognizing both its strengths and limitations—and those limitations really don’t become apparent until you use it every day.

Understanding the Beast

We are all in the early stages of this journey, so, I want to share my reality check on the strengths and weaknesses of GenAI and humans leading me to take a new approach.

Strengths

GenAIHumans
  1. Makes suggestions by digesting vast amounts of information.
  2. Instant teaching of new languages and tools replacing study.
  3. Rapidly generates good, if generic, documents, presentations, software, websites, videos, etc.
  4. Saves masses of time on tedious tasks.
  5. Today, you don’t have to be a coder to be a coder.
  1. Actual intelligence (discernment, subject matter expertise, experience, motivation).
  2. Innovative, creative with the ability to invent ideas and inspiration from pure thought.
  3. Business, technical and life lessons learned from experience that enable us to ask the question “Why.”
  4. Understanding relationships and creating emotional responses.
  5. Ability to multi-task between many interrelated real-time developments

Limitations

GenAIHumans
  1. It’s not curious – only responsive. It does not ask users proactive or “why” questions because it can’t actually “think.”
  2. Its information is based on acquired data and cannot generate innovative ideas.
  3. It only understands context if fed to it, having no implementation experience.
  4. Its input is from historical information unless constant costly relearning is in the budget. Not able to adapt to current thinking and events.
  5. It advises from frequently conflicting data and instructions for out-of-date systems because it doesn’t ask questions before suggesting.
  6. Ideas are presented as facts rather than frequently incorrect or incomplete suggestions.
  7. It overwhelms with information that users must sift through. If errors are found, GenAI apologizes, providing new suggestions, leaving users unsure if better answers were available.
  1. Users/humans can’t ask questions about “what they don’t know that they don’t know.
  2. They struggle to filter and discern masses of ideas and get overwhelmed with too much information.
  3. Inability to discriminate between facts and persuasive marketing ideas because of emotional constraints such as the fear of job loss.
  4. Understanding that most decisions have an emotional content and are governed by “political” choices
  5. Having the default way of being that is governed by remembered facts and emotions of the past that we make mean something, rather than living in the present.

Concerns

  • A recent study from Arctic Wolf revealed that AI has replaced malware as the biggest cybersecurity concern. I suspect the fear of the unknown and the opaqueness of LLM privacy are the reasons. Hopefully, this work removes some of these unknowns.
  • When an answer is incorrect, GenAI covers up. For example, my wife, Michelle, was working on a project. After two hours of GenAI not completing a task, she asked why. It said, “I’m designed to sound helpful even when I hit hard limits and that creates a gap between what I say I can deliver and what I can actually deliver.”
  • Consumption of vast amounts of power and ecological impact with optimism of a solution.

The Verdict So Far

We must adapt. The human will always be the thinker and the controller, and the AI is the accelerator. Understanding the above comparison, Generative AI does make what we do more valuable, productive and enables our personal creativity. 

Referring to Gen AI as a partner, companion, helper, useful idiot, or expensive parrot gives it human attributes implies human qualities it cannot have. 

It is remarkable software programmed to engage the user and yes it often infuriates as a trial-and error machine!

Building a Proactive, Expert-Guided Cybersecurity System

The shift in my journey was to see how to combine the strengths of both AI systems and humans. This meant defining a system that does not rely on what a generative AI “remembers.” It should draw upon combined subject matter expertise, dynamically integrated and updated with today’s thinking and Generative AI questioning into private informative guidance.

This is integrated with transformative expert recommendations, measurement and GenAI-style questioning, into a single customizable system, which I have termed an Expert-Guided Cybersecurity System. 

As with Agentic AI, this is likely one of several innovative approaches.

It combines the human aspects of current user context with AI’s ability to draw upon relevant information, taking us from overload to expertise.

The journey has been full of learning moments, such as managing the sheer volume of information—a common issue with AI responses: the dreaded “too long; didn’t read.” Providing transformative questions deals with the limitation that users don’t know which questions to ask.

Why This Matters

Cybersecurity is no longer just an IT issue—it’s a business imperative. However, most organizations don’t have the resources to sift through every standard framework or emerging threat. 

That’s where this system makes a difference. It distills expert guidance into actionable steps tailored to the user’s context and keeps evolving with new data. The good news is this approach can be applied to adjacent areas: resilience, HR, and, in fact, any guidance or white paper topic that moves from what to do to how to do it.

Happy Ending

Building this platform has been part engineering, part philosophy. It’s a recognition that generative AI, while powerful, isn’t enough. We need systems that guide, prioritize, learn transparently, and provide guidance. I have no doubt that without AI, I could not have built my new system. I hope you have found this evolving story of interest and relevant to your efforts.

The journey continues and expands on my website at cybyr.com/ai with discussions of the latest AI tools and a description of my new Expert-Guided Cybersecurity System.