Human vs AI Challenge
You answer. AI answers. See what each one missed.
It's not about who wins. Comparing your answer with an AI's shows you the points you skipped, the approaches you didn't consider, and where AI gets things wrong too.
- Step 1
Pick a question
Testing, coding, reasoning, problem solving or AI itself.
- Step 2
Answer first
Write your own answer before you see anything from the AI.
- Step 3
AI answers on its own
It gets the same question, never your answer.
- Step 4
Compare and learn
Strengths, gaps, mistakes, other approaches and a reasoned verdict.
Testing
- Medium
How would you test a login page that uses OTP (one-time password) authentication?
Trains: Test design: happy paths, edge cases, security and usability
- Medium
A UI test passes on your machine but fails about one run in five on CI. How do you find the cause and fix it?
Trains: Debugging non-deterministic failures
- Easy
List the most important test cases for an online shopping cart, and explain why each one matters.
Trains: Prioritising tests by risk
Coding
- Easy
You have a list of 10 million email addresses and need to find the duplicates. Describe your approach, its time and memory cost, and what changes if the list doesn't fit in memory.
Trains: Choosing data structures and reasoning about cost
- Hard
Design a rate limiter that allows each user 100 requests per minute across several API servers. Explain the algorithm, where state lives and how it fails.
Trains: System design trade-offs
- Medium
An API endpoint that used to respond in 200 ms now takes 4 seconds. Nothing obvious changed. Walk through how you would investigate.
Trains: Structured performance debugging
Reasoning
- Easy
A bat and a ball cost $1.10 in total. The bat costs $1.00 more than the ball. How much does the ball cost? Show your reasoning, and explain why many people get it wrong.
Trains: Checking intuitive answers
- Medium
A study of successful startups finds that 80% of their founders dropped out of college. A blog concludes that dropping out makes you more likely to succeed. What is wrong with this reasoning?
Trains: Spotting statistical traps
Problem solving
- Medium
Estimate how many smartphone repair shops a city of 2 million people needs. State your assumptions and how sensitive the answer is to each.
Trains: Estimation with explicit assumptions
- Medium
It's release day. You have 12 open bugs and time to fix 3. How do you decide which 3, and what do you do about the other 9?
Trains: Prioritising under constraints
AI literacy
- Easy
Why do large language models sometimes state false information confidently, and what can a user do to reduce the risk?
Trains: Understanding how LLMs fail
- Hard
A company wants an AI assistant that answers questions about its internal policies, which change every month. Should it fine-tune a model or use retrieval-augmented generation (RAG)? Justify your choice.
Trains: Choosing the right AI technique
- Hard
How would you test a customer-support chatbot built on a large language model, given that its answers are not the same every time?
Trains: Testing non-deterministic AI systems