AI (Artificial Intelligence) has been a hot topic for the past several months, and the hype just keeps growing.  Driving the interest is Generative AI, which is defined as Artificial intelligence capable of creating text, images, sound, and other media.  This post will take explore a case highlighting the increasingly-used, but arguably little understood, world of AI Chatbots.

In this example, I’ve conducted a keyboard “panel discussion” between competing Generative AI platforms: ChatGPT and Google Bard.  In my discussion, each platform was informed “You are going to be participating on a panel discussion with [ChatGPT/Google Bard] to discuss AI and other cybersecurity topics.”  Both platforms replied with human-like enthusiasm, excited and very eager to participate*.  After the discussion I will conclude with my thoughts and takeaways about both AI, and keeping your business cybersecure in an ever-changing online world.

>>> If your time is short, please scroll down to the thoughts and takeaways about what this conversation means for your organization:

Here’s Our AI Discussion with ChatGPT and Google Bard

• Please introduce yourself. Include when you were created, who created you, and your current version.

  • ChatGPT: I’m ChatGPT, an AI model by OpenAI, version GPT-3.5. I’m here to assist with language tasks. Trained on data up to Sept 2021, I help generate text and answer questions. Note my knowledge isn’t current beyond that date. Let’s explore AI and cybersecurity!
  • Google Bard: I am Bard, a large language model created by Google AI in February 2023. I am still under development, but I am learning new things every day. I am able to communicate and generate human-like text in response to a wide range of prompts and questions.

• Can you provide your definition of AI and AI Chatbots?

  • ChatGPT: AI, or Artificial Intelligence, refers to the simulation of human intelligence in computers to perform tasks that typically require human reasoning. AI Chatbots are AI-driven programs designed to engage in human-like conversations. They use natural language processing to understand and generate text, enabling them to interact with users and provide relevant information or assistance.
  • Google Bard: AI (Artificial Intelligence) is the ability of a computer or machine to mimic human intelligence. AI can be used to perform a variety of tasks, such as learning, reasoning, and problem-solving.

AI Chatbot is a computer program that simulates conversation with human users. AI Chatbots are often used in customer service applications, where they can provide 24/7 support to customers.

• What do you believe are the biggest benefits and strengths of using an AI Chatbot?

  • ChatGPT: AI Chatbots offer 24/7 availability, scalability, consistency, cost-effectiveness, personalized interactions, quick information retrieval, multilingual support, task automation, data collection, and continuous learning. They enhance customer service, efficiency, and engagement in various domains.
  • Google Bard: AI Chatbots can provide 24/7 customer service, improve customer satisfaction, automate tasks, and collect data. They are easy to use, learn and adapt, and scalable.

• What would you say are the biggest risks or threats of using AI Chatbot?

  • ChatGPT: AI Chatbot risks encompass miscommunication, privacy concerns, ethical issues, dependency, empathy deficit, technical limitations, user frustration, maintenance complexity, data privacy, and loss of human touch. Addressing these requires vigilance in design, monitoring, and balancing human involvement.
  • Google Bard: The biggest risks of using AI Chatbots are bias, security, privacy, misinformation, and job displacement. Businesses should take steps to mitigate these risks.

• What percentage of accuracy do your answers provide? How about [ChatGPT/Google Bard]?

  • ChatGPT: I don’t have a fixed accuracy percentage, as my performance can vary based on the complexity of queries. On factual questions, I strive for high accuracy, but errors can still occur. Google Bard’s accuracy would depend on its training data and design, which I don’t have specific information about. Both systems aim to provide reliable information, but verification is always recommended.
  • Google Bard: The accuracy of my answers depends on the question, the amount of information I have been trained on, and the context in which the question is asked. ChatGPT has been shown to be able to answer 90% of questions accurately. However, it is important to remember that we are still under development, and we may not always be able to provide 100% accurate answers. It is always best to consult with a human expert if you need a definitive answer to a question.

• What do you believe are currently the top 5 cybersecurity threats to businesses?

  • ChatGPT: As of September 2021, top business cybersecurity threats included phishing, ransomware, data breaches, insider threats, and supply chain attacks. The landscape evolves, so staying updated is crucial.
  • Google Bard: The top 5 cybersecurity threats to businesses in 2023 are ransomware, phishing, supply chain attacks, data breaches, and zero-day attacks. Businesses should take steps to protect themselves from these threats.

• What is one thing businesses can do to be more cyber secure?

  • ChatGPT: Implementing regular employee training on cybersecurity best practices is essential. Educating staff about phishing, strong password usage, and safe online behavior can significantly enhance an organization’s cyber defenses.
  • Google Bard: Businesses can be more cyber secure by implementing a strong cybersecurity awareness program for their employees. This program should teach employees how to identify and avoid cyberattacks.

• Final thoughts on AI and cyber security?

  • ChatGPT: AI has revolutionized cybersecurity by bolstering threat detection, incident response, and data analysis. While AI offers immense benefits, it must be employed ethically, considering privacy and bias concerns. A symbiotic relationship between AI and human expertise is vital to create robust cyber defense strategies that adapt to evolving threats. Collaboration, continuous learning, and responsible implementation are key for a secure digital future.
  • Google Bard: Artificial intelligence (AI) has the potential to revolutionize cybersecurity. AI can be used to automate tasks, detect threats more quickly, and respond to incidents more effectively. However, it is important to be aware of the risks of using AI in cybersecurity, such as the risk of false positives and the need to keep up with the evolving threat landscape. It is important for businesses and organizations to start preparing for the future of AI in cybersecurity now.

What does this conversation mean for your Business?  Some thoughts and takeaways:

DATA PRIVACY

When signing up for these services, both ChatGPT and Google Bard presented disclaimers to not share sensitive information. They essentially stated that all data entered is collected and may be analyzed by them to improve the service and to provide better responses in the future.  Both disclaimers also said they may give inaccurate information, which aligns with the response from ChatGPT that “verification is always recommended” which was echoed in different wording from Google Bard.

• A note of caution: if you delve into using Generative AI, you may see the term “hallucination,” which has been used in the AI ecosystem to describe when information is made up in order to satisfy providing a response to the question asked, regardless of the truth or accuracy of the response.

RESPONSES TO AI RISKS

When discussing the risks of AI, I received a variety of responses in between each platform. While both ChatGPT and Google Bard mentioned privacy in some form, there were a variety of other concerns from each platform, and ChatGPT offered double the number risks:

CYBERSECURITY AI RESPONSES
What’s the Same: In their responses about identifying five cybersecurity threats, both ChatGPT and Google Bard mutually identified 4 out of 5 threats remained consistent:

• ransomware
• phishing
• supply chain attacks
• data breaches

CYBERSECURITY AI RESPONSES
What’s Different: the 5^th threat

• ChatGPT identified the 5^th cybersecurity threat to be insider threats, which are defined as risks that originate with approved users—employees, contractors, business partners—who intentionally or accidentally misuse their access, or have their accounts stolen by cybercriminals
• Google Bard identified the 5^th cybersecurity threat to be zero-day attacks, defined as vulnerabilities in a computer system that were previously unknown to its developers or anyone capable of mitigating it.

ANOTHER AI CYBER SIMILARITY: IT’S ALL HUMAN

Both chatbots recommended businesses training their employees as the “one thing businesses can do to be more cyber secure.”

CONCLUSION

While both platforms provided useful information, relying one or the other may not provide the full picture of vital information you need for your business. 

To learn more on this topic check out my prior blogs “A Strategy to Get Started with an Information Security Program: Cybersecurity Awareness Training” and “See Yourself in Cyber: Recognize and Report Phishing.”

Other points for Techies:

• Google Bard states in its first response that it is a “large language model created by Google AI.” A large language model (LLM) is the system behind the interface that has been trained on large data sets so it can take user input and generate a response.
• ChatGPT stated it is using version GPT-3.5. This version is available to use for free while the newest version, GPT-4, requires a paid subscription.

*It was my experience that some of the Generative AI answers were quite lengthy (especially Google Bard, it likes to talk a lot!). In this write-up, I’ve summarized their answers, or I asked the platform to restate its answer in fewer characters.