LLM-as-a-Judge is an evaluation paradigm introduced by Zheng et al. (2023) that uses strong large language models (such as GPT-4) to automatically evaluate the quality of LLM-generated responses. The paper introduces two complementary benchmarks – MT-Bench and Chatbot Arena – and systematically analyzes the biases, agreement rates, and practical viability of using LLMs as scalable proxies for human evaluation.

Evaluating LLM-based chat assistants faces fundamental challenges:

• Breadth of capabilities: Models must handle writing, reasoning, coding, math, and more
• Benchmark inadequacy: Traditional NLP benchmarks (MMLU, HellaSwag) fail to capture conversational quality and human preferences
• Human evaluation cost: Obtaining reliable human preferences at scale is prohibitively expensive and slow
• Subjectivity: Open-ended tasks have no single correct answer, making automated metrics insufficient

LLM-as-a-Judge addresses these by leveraging strong LLMs to approximate human preferences at a fraction of the cost while maintaining over 80% agreement with human annotators.

MT-Bench is a multi-turn question set designed to evaluate chat assistants across 8 diverse categories:

1. Writing: Creative and professional writing tasks
2. Roleplay: Character-based conversational scenarios
3. Extraction: Information extraction from provided text
4. STEM: Science, technology, engineering, and math questions
5. Reasoning: Logical and analytical reasoning problems
6. Coding: Programming tasks and code analysis
7. Math: Mathematical computation and proof tasks
8. Humanities: History, philosophy, and social science questions

Each question has two turns – a follow-up question tests the model's ability to maintain context and build on its initial response. This multi-turn design is critical because single-turn evaluation misses crucial aspects of conversational AI quality.

Chatbot Arena is a crowdsourced battle platform where:

• Users submit open-ended prompts of their choosing
• Two anonymous LLMs generate responses side by side
• Users vote for the better response (or tie)
• Elo ratings are computed from accumulated votes

This provides a diverse, real-world evaluation signal that complements the controlled MT-Bench setting. With over 30K conversations and human preferences publicly released, it has become a de facto standard for LLM comparison.

The paper explores three LLM-as-a-Judge configurations:

• Pairwise comparison: The judge sees a question and two responses, then selects the better one or declares a tie
• Single answer grading: The judge scores a single response on a numeric scale (e.g., 1-10) with a structured rubric
• Reference-guided grading: The judge is provided a reference answer (useful for math/coding tasks with verifiable solutions)

Three systematic biases were identified and analyzed:

Position Bias LLM judges tend to favor responses presented in certain positions (e.g., the first response). GPT-4 shows high consistency across position swaps, but weaker models exhibit significant position bias.

<latex> \text{Consistency} = P(\text{same judgment} \mid \text{position swap}) </latex>

GPT-4 achieves the highest consistency rate among tested judges.

Verbosity Bias LLM judges tend to prefer longer, more detailed responses even when the additional content does not improve quality. This can unfairly penalize concise but correct answers.

Self-Enhancement Bias LLMs show a tendency to favor their own generated outputs – a model used as a judge may rate its own responses higher than those from competitors.

• Position swapping: Run evaluation twice with response order reversed, take majority vote
• Chain-of-thought prompting: Require the judge to explain reasoning before scoring, improving accuracy on math/reasoning tasks
• Reference-guided grading: Provide ground-truth answers for verifiable tasks to anchor judgment
• Few-shot examples: Include example judgments to calibrate the judge's scoring

The central finding is that strong LLM judges achieve human-level agreement:

• GPT-4 as judge: Over 80% agreement with both controlled (MT-Bench) and crowdsourced (Chatbot Arena) human preferences
• This matches the human-human agreement rate – human annotators agree with each other at approximately the same 80% level
• Agreement holds across all 8 MT-Bench categories, though math and reasoning show slightly lower concordance

import openai
 
def llm_judge_pairwise(question, response_a, response_b, model="gpt-4"):
    """Use LLM-as-a-Judge for pairwise comparison."""
    prompt = (
        "Please act as an impartial judge and evaluate the quality "
        "of the responses provided by two AI assistants to the user "
        "question below.\n\n"
        "Avoid position bias -- evaluate based on content quality only.\n\n"
        f"[User Question]\n{question}\n\n"
        f"[Assistant A]\n{response_a}\n\n"
        f"[Assistant B]\n{response_b}\n\n"
        "[Evaluation]\n"
        "Provide reasoning, then verdict: "
        '"[[A]]" if A is better, "[[B]]" if B is better, "[[C]]" for tie.'
    )
    response = openai.chat.completions.create(
        model=model,
        messages=[{"role": "user", "content": prompt}],
        temperature=0,
    )
    judgment = response.choices[0].message.content
    if "[[A]]" in judgment:
        return "A", judgment
    elif "[[B]]" in judgment:
        return "B", judgment
    return "tie", judgment
 
def evaluate_with_position_swap(question, resp_a, resp_b):
    """Mitigate position bias by evaluating in both orders."""
    verdict_1, _ = llm_judge_pairwise(question, resp_a, resp_b)
    verdict_2, _ = llm_judge_pairwise(question, resp_b, resp_a)
    # Reconcile verdicts accounting for the swap
    v2_adjusted = "B" if verdict_2 == "A" else ("A" if verdict_2 == "B" else "tie")
    if verdict_1 == v2_adjusted:
        return verdict_1
    return "tie"  # Disagreement defaults to tie

LLM-as-a-Judge has become a foundational methodology in LLM evaluation:

• Chatbot Arena has grown into the most widely referenced LLM leaderboard
• MT-Bench scores are standard in model release announcements
• The methodology is used in training pipelines (e.g., RLAIF – RL from AI Feedback)
• Frameworks like AlpacaEval and WildBench build on these principles

• Judge quality is bounded by the evaluating model's capabilities
• Biases (position, verbosity, self-enhancement) persist even with mitigation
• Weak reasoning ability limits accuracy on math and formal logic tasks
• Cultural and linguistic biases from training data affect judgment
• Cannot replace human evaluation for safety-critical applications

• Zheng et al. (2023) - Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena
• Official MT-Bench / LLM Judge Code (LMSYS)
• Chatbot Arena Live Platform
• Li et al. (2023) - AlpacaEval: Automatic Evaluator for Instruction-Following Models

• Self-Play Fine-Tuning (SPIN) - Training method evaluated using MT-Bench
• AgentBench - Benchmark for evaluating LLM agents
• tau-bench - Benchmark using database state comparison instead of LLM judgment