All traces

AI use in workforce analytics introduces substantive risks including algorithmic opacity, automation bias, proxy-based discrimination, and employee surveillance.

Practitioners reported tensions that AI assistance could erode professional judgment, reflective value of writing, and create deskilling risk and ethical complexity.

Adoption of blockchain and AI in accounting faces scalability, interoperability, and potential job displacement concerns that constrain effective integration.

Most agentic AI studies in healthcare were exploratory, limited in scope, lacked robust clinical validation, and lacked conceptual clarity, with only one trial involving patients.

High religiosity marginally weakens the positive link between digitalization and MSME labor productivity, reflecting transitional adaptation frictions in highly normative environments.

Higher social-interaction burnout and subjective loneliness predict stronger emotional attachment to AI companions among young adults, with parasocial interaction mediating the relationship.

Autonomous research systems could overwhelm peer review infrastructure and pollute scientific literature with low-quality or noisy papers.

Relying on passive AI copying at work reduced self-efficacy, psychological ownership, and work meaningfulness, with declines in efficacy and meaningfulness persisting after returning to manual work.

Local journalists in Germany do not fully leverage AI to support data-related reporting work, linked to limited awareness of what AI can do.

Rapid proliferation of AI-mediated digital afterlife technologies including chatbots trained on personal data, voice clones and posthumous avatars has created moral risks of posthumous simulation without operational governance constraints.

95 listeners could barely distinguish natural speech from ElevenLabs-cloned speech after telecom transmission, with overall accuracy of 54.8% and only 44.0% on VoLTE, increasing susceptibility to voice spoofing.

Detection skill did not generalize across modalities, and confidence did not track voice accuracy, leaving people vulnerable to deepfake-enabled identity theft and financial fraud despite some detection ability.

Same preservice teachers raised concerns that LLM outputs contained inaccuracies, encouraged overreliance, and were less useful for validation phases of modeling.

An AI-generated imitation of Drake and The Weeknd's voices in a viral track convinced millions of listeners it was authentic, requiring a label disclosure.

Current AI studies for sport-related concussion are frequently constrained by small or imbalanced samples, inconsistent definitions, limited external validation, and poor model interpretability.