As of 13 August 2026, AI can only partly score interview answers against a rubric.

What to actually do

1. Hand it to a person

   The route this page recommends

   A person who owns the outcome does this end to end, worth it when the failure is dear.

2. Use a tool built for this

   Second choice

3. Do it yourself

   The distant third

   A chat interface, chat-fluent skill, and roughly 30 minutes until you can act on the result.

   How to actually do it

   1. Open the approved interview rubric and question set, and confirm that each criterion has clear descriptions for every score used.
   2. Remove unnecessary names and personal details from the interview transcript, then check that you have permission to process the candidate data with the chosen AI service.
   3. Paste the role context, questions, rubric and one anonymised candidate's answers into a chatbot using the prompt above.
   4. Ask the model to score each answer separately and to quote the answer evidence rather than filling gaps with assumptions.
   5. Copy the score table into the hiring record and compare every score with the rubric and the quoted answer evidence, changing any score that does not meet the stated standard.
   6. Have a second interviewer or HR colleague check consistency, bias risks and any flagged criterion before the panel makes its own documented decision.

   Prompt

   Score the interview answers below against the supplied rubric for the role. Use only evidence in each answer and do not infer personality, protected characteristics, health, family circumstances, nationality, age or other personal traits. For each question, give: the criterion being assessed, a score from 1 to 5, the exact evidence supporting the score, what evidence is missing, and a short explanation. Apply the same standard to every candidate. If an answer is ambiguous, say so instead of filling in the gap. Do not rank candidates, recommend rejection or selection, or make a hiring decision. Flag any criterion that is unclear, potentially discriminatory or not evidenced by the question. End with a table showing each score and a list of points that a human interviewer must check.
   
   Role and assessment context:
   [PASTE ROLE, QUESTION SET AND SCORING INSTRUCTIONS]
   
   Rubric:
   [PASTE THE APPROVED RUBRIC, INCLUDING WHAT SCORES 1 TO 5 MEAN]
   
   Candidate identifier:
   [USE AN ANONYMISED ID]
   
   Interview answers:
   [PASTE THE ANSWERS, WITH PERSONAL DETAILS REMOVED WHERE THEY ARE NOT NEEDED]
   
   Return the assessment in plain English and do not invent facts.

   Open it prefilled in ChatGPT or Claude, or copy it into Gemini, which takes no prefill link.