All traces

Current newsroom AI assessments focus narrowly on technical accuracy or efficiency and struggle to capture broader organizational and ethical implications for trust, governance, and human collaboration.

In the matched Stage II-III cohort, patients with low lymph node ratio showed potential harm from adjuvant chemotherapy, indicating heterogeneous treatment effects and risk of overtreatment.

AI auto-contouring raises concerns that future radiation therapy practitioners may have reduced ability to critically evaluate auto-generated contours.

The TyG-FI showed limited incremental predictive advantage over FI-Lab alone and reduced discrimination on external validation, with AUROCs dropping to 0.694 and 0.667 in eICU compared to 0.764 and 0.769 internally.

Sentinel-2 and CART-based operational AGB mapping in Pinus brutia produced total estimates that differed by more than 700,000 Mg across the same 13,687 ha area depending solely on which allometric reference was used, making reference selection a dominant source of uncertainty.

Technostress mediated the relationship, such that when openness to change and positive AI attitudes were low, higher technostress was associated with reduced innovative work behavior among the same hospital staff.

In educational use, ChatGPT can generate wrong information and reflect training-data biases that may augment existing biases, raising privacy issues.

Multi-omics biomarker approaches face persistent challenges in data heterogeneity, reproducibility, and clinical validation across diverse patient populations.

Oncology AI systems can reproduce or amplify existing disparities across patient populations, and efforts to enforce fairness definitions often conflict with overall predictive performance.

When used for prostate MRI reporting, LLMs can hallucinate measurements, flip negations, misstate laterality, and overstate cancer likelihood, creating patient-safety and accountability risks especially if reports are copied outside clinical governance.

Perceived risk around AIGC reduced fashion designers' feelings of autonomy, competence and relatedness, undermining psychological conditions for adoption.

Most models lacked external validation, only one was low risk of bias, publication bias was detected, and performance dropped in HIV-positive populations, leaving models not ready for routine clinical implementation.

AI systems applied to healthcare decision-making, medical diagnosis, and other domains can lead to unfair outcomes that perpetuate existing inequalities and reinforce harmful stereotypes, including generative biases in synthetic media.

Artificial intelligence approaches for CT-based prediction of hematoma expansion and adverse outcomes after spontaneous ICH are limited by small cohorts, overfitting, dataset heterogeneity, insufficient external validation, poor interpretability and lack of workflow integration.

Despite belief AI will revolutionise nursing education, actual implementation remains conservative at augmentation level with none achieving transformative redefinition.