Responsible Use of AI in Testlify Assessments
Where AI may be used
Depending on the product and configuration, Testlify may use AI-assisted capabilities in areas such as:
- assessment-content support
- response scoring or classification
- report or insight generation
- interview support
- proctoring and integrity signals
- workflow automation
The exact use of AI depends on the feature selected.
Our responsible-use principles
AI-assisted assessment features should be evaluated against the same core expectations as other employment assessment methods.
These include:
- job relevance
- validity for the intended use
- consistency
- fairness
- accessibility
- transparency
- privacy and security
- human oversight
- documentation
- ongoing monitoring
AI does not remove the need for assessment science or responsible decision-making.
Is AI the final hiring decision-maker?
Testlify provides evidence and decision-support tools. The employer remains responsible for the employment decision.
Customers should not rely on an AI-generated score, summary, flag, or recommendation as the sole basis for a high-stakes decision without appropriate review.
How should AI-generated insights be interpreted?
AI-generated insights can help organize or summarize evidence. They may be incomplete or incorrect and should be checked against:
- the candidate's actual response
- defined scoring criteria
- job requirements
- other assessment evidence
- human reviewer judgment
How does Testlify approach AI fairness?
Fairness requires more than removing protected characteristics from a model.
Relevant considerations include:
- whether the input data is job-related
- whether the scoring method measures the intended construct
- subgroup performance
- accessibility
- explainability
- monitoring for drift or unexpected outcomes
- the availability of human review
No AI system should be described as completely unbiased.
Frequently asked questions
Does Testlify use AI to automatically hire or reject candidates?
Testlify supports assessment and decision workflows. Employers control their hiring decisions and should maintain appropriate human oversight.
Are AI scores always accurate?
No. AI-assisted outputs should be evaluated, monitored, and interpreted within their intended context.
Can customers review the basis of an AI-generated result?
The available level of explanation depends on the feature. Customers should request documentation for high-stakes uses.
Does AI remove bias?
No. AI may reduce some forms of inconsistency while introducing or reproducing others.
Should proctoring AI flags lead to automatic rejection?
No. A flag should be treated as a signal for contextual review.