Build your judgement.
Question the response.
Learn to spot mistakes, weigh evidence and explain what makes an AI answer useful. Short exercises. Clear feedback. Room to think.
Free to use · No account needed · No coding required
Does the answer match the evidence?
In a survey, 18 of 30 people said they preferred the new design.
“The survey shows that 90% of people preferred the new design.”
Look beyond the confident wording. Check the numbers and what the survey actually establishes.
Different models. The same need for careful judgement.
Practise skills you can use with any AI. The lab is independent of the featured AI services.
ChatGPT
Claude
Gemini
Grok
Llama
DeepSeek
Qwen
Mistral
128practice exercises
3difficulty levels
5 or 10questions in a short session
Six ways to sharpen your judgement.
Begin with the skill you want to practise. Most exercises are untimed, so you can take a careful look.
Evaluate a response
Use a clear rubric to judge accuracy, relevance, instructions and clarity.
Explore this skillCompare two answers
Decide which response best serves the user—and explain why.
Explore this skillFollow the instructions
Notice the small constraints that can change the whole task.
Explore this skillCheck the evidence
Separate supported claims from assumptions and uncertainty.
Explore this skillRecognise bias
Challenge the first explanation and consider alternatives.
Explore this skillPay attention to detail
Practise precise reading in an eight-minute attention session.
Explore this skillRead. Judge. Reflect.
You don’t need to use a particular AI model. The prompts and responses are provided for you.
- 1
Read the prompt and response
Check what was asked, what was said and what evidence is available.
- 2
Make your judgement
Choose an answer or apply a rubric. Record your confidence and, if you like, your reasoning.
- 3
Learn from the feedback
Review the answer key, revisit mistakes and see how your first attempts develop over time.
Start with five questions.
Choose a mixed session and your difficulty level. Take your time, then review what you learned.
About the Project
AI Response Lab is an independent project created by Leigh Shellam to explore and practise structured human evaluation of AI-generated responses.
The Lab is designed around the skills involved in assessing AI output carefully and consistently, with an emphasis on evidence, reasoning, instruction-following, uncertainty and clear justification of evaluation decisions.
What the Lab Develops
- Applying evaluation criteria consistently
- Comparing competing AI responses
- Identifying unsupported claims and reasoning gaps
- Assessing instruction-following and relevance
- Recognising bias, ambiguity and uncertainty
- Explaining evaluation decisions clearly and objectively
Independent Project
AI Response Lab is an independent educational project created by Leigh Shellam. It is not affiliated with, sponsored by, or endorsed by OpenAI, Anthropic, Google or any other AI provider.