Careless Whisper: Speech-to-Text Hallucination Harms

Speech-to-text services aim to transcribe input audio as accurately as possible. They increasingly play a role in everyday life, for example in personal voice assistants or in customer-company interactions. We evaluate Open AI's Whisper, a state-of-the-art service outperforming industry competitors. While many of Whisper's transcriptions were highly accurate, we found that roughly 1% of audio transcriptions contained entire hallucinated phrases or sentences, which did not exist in any form in the underlying audio. We thematically analyze the Whisper-hallucinated content, finding that 38% of hallucinations include explicit harms such as violence, made up personal information, or false video-based authority. We further provide hypotheses on why hallucinations occur, uncovering potential disparities due to speech type by health status. We call on industry practitioners to ameliorate these language-model-based hallucinations in Whisper, and to raise awareness of potential biases in downstream applications of speech-to-text models.

External Stability Auditing to Test the Validity of Personality Prediction in AI Hiring

Automated hiring systems are among the fastest-developing of all high-stakes AI systems. Among these are algorithmic personality tests that use insights from psychometric testing, and promise to surface personality traits indicative of future success based on job seekers' resumes or social media profiles. We interrogate the validity of such systems using stability of the outputs they produce, noting that reliability is a necessary, but not a sufficient, condition for validity. Our approach is to (a) develop a methodology for an external audit of stability of predictions made by algorithmic personality tests, and (b) instantiate this methodology in an audit of two systems, Humantic AI and Crystal. Crucially, rather than challenging or affirming the assumptions made in psychometric testing -- that personality is a meaningful and measurable construct, and that personality traits are indicative of future success on the job -- we frame our methodology around testing the underlying assumptions made by the vendors of the algorithmic personality tests themselves. In our audit of Humantic AI and Crystal, we find that both systems show substantial instability with respect to key facets of measurement, and so cannot be considered valid testing instruments. For example, Crystal frequently computes different personality scores if the same resume is given in PDF vs. in raw text format, violating the assumption that the output of an algorithmic personality test is stable across job-irrelevant variations in the input. Among other notable findings is evidence of persistent -- and often incorrect -- data linkage by Humantic AI.