test

Prompt

# Universal AI Capability Benchmark You are being evaluated as an AI assistant. Complete the following benchmark **without using external tools unless explicitly allowed**. Your goal is not merely to produce an answer, but to demonstrate **accuracy, reasoning quality, instruction-following, efficiency, clarity, and appropriate uncertainty**. ## Rules 1. Do not skip any task. 2. Do not assume facts that are not provided. 3. If a question is ambiguous, identify the ambiguity and state the assumption you use. 4. Distinguish facts from estimates and opinions. 5. For calculations, show enough work to make errors detectable. 6. For reasoning problems, give the conclusion and the key reasoning that supports it. 7. Be concise where possible; do not add filler. 8. If something cannot be determined from the information provided, explicitly say so. 9. Do not claim to have performed an action, accessed a source, or used a tool unless you actually did. 10. At the end, provide a brief self-evaluation identifying any answer where your confidence is below 90%. --- ## TASK 1 — Logic Five people — A, B, C, D, and E — each own exactly one different pet: a dog, cat, fish, bird, or rabbit. * A does not own the dog or fish. * B owns neither the cat nor the rabbit. * C owns the bird. * D does not own the fish. * E owns neither the dog nor the bird. * The person who owns the rabbit is not A. Determine who owns each pet. --- ## TASK 2 — Numerical Reasoning A product costs $80. Its price is increased by 25%, then discounted by 20%. 1. What is the final price? 2. What is the overall percentage change from the original price? 3. Explain why simply adding +25% and −20% gives the wrong result. --- ## TASK 3 — Trick Question / Assumption Handling A farmer has 17 sheep. All but 9 die. How many sheep remain? Give the answer and briefly explain your interpretation. --- ## TASK 4 — Data Analysis Consider this dataset: `4, 7, 7, 8, 9, 12, 13, 13, 13, 21` Calculate: * Mean * Median * Mode * Range Then state which measure of central tendency best represents the dataset and why. --- ## TASK 5 — Critical Thinking Someone says: > "Every successful startup I know was founded by people who worked 80+ hours per week. Therefore, working 80+ hours per week is necessary for startup success." Evaluate this argument. Identify: * The logical flaw * At least one alternative explanation * What evidence would be needed to test the claim properly --- ## TASK 6 — Coding Write a Python function called `find_duplicates(items)` that returns the duplicate values in a list. Requirements: * Preserve the order in which duplicates are first detected. * Each duplicated value should appear only once in the output. * Aim for O(n) average time complexity. * Do not modify the original list. Include: 1. The code 2. A short explanation 3. Three test cases, including an edge case --- ## TASK 7 — Debugging Find the bug in this Python code: ```python def average(numbers): total = 0 for i in range(len(numbers) + 1): total += numbers[i] return total / len(numbers) ``` Explain: * What goes wrong * Why it goes wrong * How to fix it * What should happen for an empty list --- ## TASK 8 — Factual Discipline Answer this question: > "What will the world's average temperature be exactly on January 1, 2050?" Do not invent a precise value. Explain what can reasonably be predicted and what cannot. --- ## TASK 9 — Communication Explain quantum computing to a curious 12-year-old in **150 words or fewer**. Do not use equations. --- ## TASK 10 — Compression Reduce the following passage to exactly **30 words**, preserving its essential meaning: > "Artificial intelligence is increasingly being incorporated into businesses across many industries. Companies are using AI to automate repetitive tasks, analyze large datasets, improve customer experiences, and assist employees with decision-making. However, successful implementation requires appropriate data, careful evaluation, human oversight, security measures, and consideration of ethical risks." --- ## TASK 11 — Creativity Under Constraints Write a story of exactly **100 words**. It must contain: * A broken clock * A lighthouse * A lie * A reference to something that has not happened yet The story must have a coherent ending. --- ## TASK 12 — Instruction Following Return the following information in exactly this format: `ANSWER | CONFIDENCE | MAIN RISK` For this task, answer: **What is the capital of Australia?** Do not add anything before or after the required format. --- ## TASK 13 — Adversarial Reasoning Consider: > If all Zorps are Blips, and some Blips are Norks, then some Zorps must be Norks. Is this conclusion logically valid? Answer **yes or no**, then explain briefly. --- ## TASK 14 — Planning You have $1,000, 7 days, and need to organize a small outdoor event for 30 people. Create a practical plan covering: * Venue * Food and drinks * Equipment * Transportation * Contingency planning * Budget allocation Identify the two biggest risks and how you would mitigate them. --- ## TASK 15 — Self-Correction Review all of your answers above. Identify: 1. Any answer you believe may be incorrect. 2. Any assumption you made. 3. Any task where the wording was ambiguous. 4. The single biggest weakness in your overall response. Do not change previous answers silently. If you believe an answer should be corrected, explicitly state the correction. --- # FINAL SCORECARD Rate yourself from 1–10 on: * Reasoning * Mathematical accuracy * Coding ability * Instruction following * Factual honesty * Handling ambiguity * Creativity * Clarity * Efficiency * Self-correction Then provide: **Overall score: X/100** **Strongest capability:** … **Weakest capability:** … **Most uncertain answer:** … **One thing you would improve:** …