220-1102 Question 823
Single answerHallucinationsA help desk technician is testing a new AI chatbot that the company wants to use for internal IT support. During a pilot, an employee asks the chatbot for steps to reset a Windows account password in the company's environment. The chatbot responds with a detailed procedure that references tools and menu paths that do not exist in the company's systems. The technician needs to explain the issue to management and recommend the best immediate mitigation. Which option BEST describes what happened and the most appropriate response?
- A
The chatbot experienced a hallucination; staff should verify AI-generated instructions against approved company documentation before following them.
- B
The chatbot encountered a permissions error; granting it administrative access will allow it to provide accurate support steps.
- C
The chatbot was affected by network latency; retrying the same prompt later will ensure the response matches the company's environment.
- D
The chatbot failed because the employee used natural language; users should submit only command-line syntax to get reliable answers.
Show answer and explanation
Correct answer: A
Explanation
This scenario describes an AI hallucination: the chatbot produced confident, detailed instructions that were not grounded in the organization's actual tools or processes. For A+ Core 2, the practical takeaway is that AI-assisted support can be useful, but technicians must validate outputs before using them in production or sharing them with end users. A best-practice response is to cross-check AI-generated answers against internal IT documentation, change-management records, and official vendor resources such as Microsoft support documentation for Windows administrative procedures. In security and support contexts, unverified AI instructions can lead to account lockouts, misconfiguration, privilege issues, or policy violations. The immediate mitigation is not more access or different prompt style, but human verification and use of approved documentation.
- A. Correct.
Correct. In AI contexts, a hallucination is when the system generates plausible-sounding but incorrect or fabricated information. The safest immediate mitigation in an IT support environment is to require verification against trusted internal knowledge bases, standard operating procedures, vendor documentation, or other approved sources before acting on the AI's output.
- B. Incorrect.
Incorrect. A permissions problem would not normally cause an AI system to invent non-existent tools or menu paths. Giving broader administrative access is also a poor immediate response because it increases risk without addressing the underlying issue of inaccurate generated content.
- C. Incorrect.
Incorrect. Network latency may delay a response, but it does not explain fabricated technical steps. Retrying might produce a different answer, but it does not guarantee accuracy. The core problem is unverified and potentially false content, not response timing.
- D. Incorrect.
Incorrect. Modern chatbots are designed to accept natural-language prompts. Requiring command-line syntax does not solve the risk of hallucinated content. The issue is the model producing incorrect information, not the conversational format of the request.