Official Question UPSC 2026
Which of the following statements with regard to Large Language Models (LLMs) used in machine learning is/are correct ? 1. LLMs assign probabilities to the next possible words and then pick the one with the highest probability. 2. LLMs process data through mathematical optimization to minimise prediction errors. 3. LLMs produce unbiased outputs. Select the answer using the code given below :
Solution & Analysis
Answer: B
To understand AI, we must look at foundational Machine Learning mechanics:
•Statement 1 is Correct: The core functional mechanism of an LLM (like GPT) is text prediction. It analyzes a context string and assigns a statistical probability distribution to all possible next words (tokens). In its most fundamental state (greedy decoding), it simply picks the word mathematically assigned the highest probability to generate text.
•Statement 2 is Correct: The training of an LLM relies on rigorous mathematical optimization. Utilizing algorithms like Stochastic Gradient Descent, the model continuously adjusts billions of internal parameters to minimize a 'loss function'—a mathematical representation of its prediction errors compared to the actual training data.
•Statement 3 is Incorrect: LLMs are strictly pattern-matching engines; they do not possess objective morality. Because they are trained on massive datasets scraped from human-generated internet text, they inherently absorb, reflect, and often amplify human prejudices, resulting in outputs that are frequently heavily biased.
Conclusion: Statements 1 and 2 represent accurate computational science, making Option B correct.