How to Use AI Confidence Scores in interviewstream
In this lesson, we’ll show you how to read and interpret AI Confidence Scores which can be included with the interview summaries generated by interviewstream’s AI Interview Summary tool. Confidence Scores give you a quick view into how accurately the AI summary reflects the candidate’s actual response, helping your team review smarter and make more informed hiring decisions.
Please Note: AI Confidence Scores are available only for interviews completed after the AI Interview Summary tool with Confidence Scoring has been activated in your account. If you have not already done so, please review the Enabling AI Features within a Requisition lesson before beginning. If you do not see confidence scores, reach out to your Customer Success Manager or contact us here.
- Navigate to a completed candidate interview record and click the Interview Summary tab to view the full AI-generated summary.
- Each summary (overall and per question) will display a Confidence Score, represented as a percentage out of 100. This score reflects how closely the summary aligns with the candidate’s full response.
- Click the Confidence Score section to open up the scoring details.
- Confidence Scores are based on multiple AI analysis factors, including:
- Textual Overlap: How much the AI summary language shares in common with the language used in the candidate’s spoken response
- Key Concept Match: Whether the core topics and ideas were accurately captured
- Factual Consistency: How consistent the claims made by the candidate are reflected in the response
- Gaps Analysis: Any critical information that may be missing from the summary
- Confidence Scores are based on multiple AI analysis factors, including:
- Confidence Scores can help standardize how your team approaches review, ensuring everyone has clear insight into the reliability of each summary before making a decision.
To export summaries with Confidence Scores, click the export button at the top of the interview record. A downloadable PDF will include both summary content and associated confidence insights.
