10 questions
Explanation: LLM means Large Language Model, the technology behind chat-based drafting assistants. The other three are invented expansions.
Explanation: Machine translation can distort terms such as suspension, sanction or deemed, which carry a specific administrative meaning. For example sanction in the sense of approval may be rendered as a restriction, which changes the whole effect of the notice.
Explanation: Abstractive summarisation rewrites the content in new words, while extractive summarisation selects and joins original sentences. Because rewriting can shift the meaning, the original must still be read before a decision.
Explanation: Transcription, also called speech to text or ASR, converts spoken audio into text. Translation changes the language, text-to-speech works in the opposite direction, and OCR applies to printed or scanned text.
Explanation: A hallucination is fluent, confident output that is factually wrong or invented, such as a non-existent rule number or a fake order date. A crash is a technical failure and a refusal is a deliberate response, so neither counts.
Explanation: Statement 1 is false, because the popularity of a tool does not remove the need to verify. Statements 2 and 3 are correct: AI can hallucinate references, and accountability always stays with the signing officer. Rejecting statement 1 removes three options at once.
Explanation: A prompt is the user input instruction, and a clear role, context, task, format and constraints produce better output. A token is the word fragment an LLM processes, while scripts and macros are automation programs.
Explanation: The Digital Personal Data Protection Act, 2023 governs personal data in India and applies when official records containing personal particulars are handled through any tool, including AI tools.
Explanation: BHASHINI is the National Language Translation Mission under MeitY. Because the subject matter is language, candidates often pick Education or Culture by mistake.
Explanation: Confidential and personal data must not go to unapproved public tools. Deleting the chat does not help, because the data has already reached an outside server, and forwarding it shares a confidential record with an unauthorised person, which may also breach the conduct rules.
Back to Common AI Tools for Administration