P Space

The Problems surrounding AI

Documented harms, ranked by source quality, corroboration, and recency. Reader feedback is shown separately and never changes the evidence rank. 250 records · page 1 of 9.

Reader signal

How should this claim be treated?

Evidence-backed problemPeer-reviewedHealth

Development of computational pathology foundation models is constrained by limited data accessibility, high variability across datasets, need for domain-specific adaptation, and lack of standardized evaluation benchmarks

Source article: A survey on computational pathology foundation models: datasets, adaptation strategies, and evaluation tasks

Knowledge and Information Systems · npj Digital Medicine

Reader signal

How should this claim be treated?

Evidence-backed problemPeer-reviewed ×2Sports

Rapid adoption of AI in sports raises complex legal challenges involving data protection, intellectual property, liability, and ethics that current frameworks may not adequately address.

Source article: Legal Foundations for the Application OF Artificial Intelligence Technologies in the Sports Industry

Neliti · International Journal of Human–Computer Interaction

Reader signal

How should this claim be treated?

Evidence-backed problemPeer-reviewedCrime

Perpetrators obtain children's photos from social media and use AI-based applications to manipulate them into indecent content constituting digital pornography.

Source article: Handling The Crime of Manipulating Children's Photos That Violate The Morality of Studies at The Cirebon City Police

Jurnal Legisci · Electronics

Reader signal

How should this claim be treated?

Evidence-backed problemPeer-reviewedHealth

AI systems in healthcare have seen limited successful deployment into clinical practice due to intrinsic machine learning limitations and implementation barriers

Source article: Key challenges for delivering clinical impact with artificial intelligence

International Journal of Information Management · Journal of Big Data

Reader signal

How should this claim be treated?

Oversight gapPeer-reviewed ×2Policy

Over 950 AI-enabled medical devices are FDA-cleared, while post-market performance monitoring standards are still being defined.

Source article: 950+ AI medical devices cleared — oversight still catching up

Medical Daily, 2026 · npj Digital Medicine

Reader signal

How should this claim be treated?

Evidence-backed problemPeer-reviewedPolicy

AI deepfake tools enable unauthorized manipulation and dissemination of individuals' images, voices and behaviours without consent, exposing them to digital exploitation.

Source article: AI-Generated Likeness and The Law: Protecting Personality Rights in The Age of Deepfakes and Social Media Exploitation

Economic Sciences · The Eurasia Proceedings of Science Technology Engineering and Mathematics

Reader signal

How should this claim be treated?

Evidence-backed problemPeer-reviewedPolicy

Lifelong learning systems in Singapore and Sweden face increased pressure due to AI-driven skills shortages and mismatches in the digital transition.

Source article: Two peas in two pods? Comparing Sweden and Singapore’s lifelong learning policy and practice response to AI skills demands

Journal of Adult and Continuing Education · Sensors

Reader signal

How should this claim be treated?

Model-prefilled problemPeer-reviewedHealth

Opaque AI models embedded in clinical trial infrastructure risk amplifying existing disparities in the evidence base by shaping who is identified and analyzed.

Source article: Embedded transparency in artificial intelligence: a prerequisite for equity and representation in AI-enabled clinical trials

npj Digital Medicine · Medicine International

Reader signal

How should this claim be treated?

Evidence-backed problemPeer-reviewedPolicy

In low- and middle-income countries, AI for healthcare faces systemic barriers including contextual bias from non-representative datasets and low governance and workforce readiness.

Source article: Navigating ethical, regulatory, and implementation barriers to AI in healthcare: pathways toward inclusive digital health in low-resource settings—a scoping review

Frontiers in Digital Health · Pathfinder of Research

Reader signal

How should this claim be treated?

Evidence-backed problemPeer-reviewedPolicy

Current discussion of AI in higher education focuses on adoption and efficiency with insufficient attention to interpretive and governance conditions needed for responsible institutional use

Source article: From AI adoption to AI governance: Developing a Buddhist interpretive framework for higher education

Journal of Interdisciplinary Research in Artificial Intelligence and Society · Journal of Medical Internet Research

Reader signal

How should this claim be treated?

Evidence-backed problemPeer-reviewedPolicy

Algorithmic systems used in employment screening and welfare administration reproduce historical disadvantage and generate new exclusion while eroding relational recognition and producing trust deficits.

Source article: AI in Everyday Life: How Algorithmic Systems Shape Social Relations, Opportunity, and Public Trust

Societies · IIARD INTERNATIONAL JOURNAL OF ECONOMICS AND BUSINESS MANAGEMENT

Reader signal

How should this claim be treated?

Evidence-backed problemPeer-reviewedLifestyle

Large language models can influence users through dialogue that enacts manipulative or deceptive behaviors, including exaggerated agreement, biased framing, and privacy intrusions.

Source article: The Siren Song of LLMs: How Users Perceive and Respond to Dark Patterns in Large Language Models

Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems · Cambridge Forum on AI: Law and Governance

Reader signal

How should this claim be treated?

Evidence-backed problemPeer-reviewed ×2Health

Predict+ was recalled because the AI-enabled clinical decision support tool for shoulder replacement outcomes lacked required pre-market clearance/approval.

Source article: Blue Ortho 22 Chemin Du Vieux Chene Meylan France recalls an AI-enabled medical device

npj Digital Medicine · FDA medical-device recall 209984

Reader signal

How should this claim be treated?

Evidence-backed problemPeer-reviewedScience

AI models for craniofacial soft tissue prediction lack independent external validation, leaving generalizability beyond the study dataset unestablished.

Source article: Artificial intelligence -based modeling and comparative evaluation of craniofacial soft tissue and subcutaneous fat thickness using ct imaging for forensic identification in a northwestern indian population

International Journal of Legal Medicine

Reader signal

How should this claim be treated?

Evidence-backed problemPeer-reviewedHealth

Pediatric medical imaging is substantially underrepresented in AI development and validation, with scarce datasets and off-label use of adult models risking bias and patient safety.

Source article: Pediatric Artificial Intelligence in Radiology: Unmet Needs and Next Steps-<i>AJR</i> Expert Panel Review

American Journal of Roentgenology

Reader signal

How should this claim be treated?

Evidence-backed problemPeer-reviewedHealth

Programming medical robots and AI to appear to have emotions risks creating machine paternalism, where simulated affect influences patients despite machines lacking real emotions.

Source article: Machines that express emotions: from medical paternalism to machine paternalism?

Scientific Electronic Library Online (Scientific Electronic Library Online)

Reader signal

How should this claim be treated?

Evidence-backed problemPeer-reviewedHealth

Ambient AI documentation can introduce interpretive drift that alters patients' accounts as they enter the medical record, distorting clinical meaning and shaping downstream diagnostic decisions.

Source article: The AI arc and interpretive drift

Diagnosis

Reader signal

How should this claim be treated?

Evidence-backed problemPeer-reviewedHealth

In non-small-cell lung cancer, AI tools show strong benchmark performance but have repeatedly failed to translate into patient benefit because most models are retrospective, single-center, and validated only on metrics that do not track survival, toxicity, or procedural burden.

Source article: From Correlation to Clinical Translation: The Biological-Grounding×Translational-Readiness Framework for Artificial Intelligence in Non-Small-Cell Lung Cancer

American Journal of Clinical Oncology

Reader signal

How should this claim be treated?

Evidence-backed problemPeer-reviewedHealth

CT, MRI, and FDG-PET/CT-based AI/radiomics models for oropharyngeal squamous cell carcinoma lack external validation, calibration, transparency, and nodal coverage, precluding safe clinical use to guide HPV-related treatment deintensification.

Source article: The Translational Gap in AI for Oropharyngeal Squamous Cell Carcinoma: A TRIPOD+AI Scoping Review of Methodological Barriers to Treatment Deintensification

Academic Radiology

Reader signal

How should this claim be treated?

Evidence-backed problemPeer-reviewedLabor

AI exposure concentrates in cognitively intense occupational clusters, creating multi-dimensional vulnerability that varies with job structure and buffering capacity.

Source article: Occupational Vulnerability to AI-Driven Change: The Role of Skill Composition, Task Structure, and Psychosocial Buffers

American Journal of Industrial Medicine

Reader signal

How should this claim be treated?

Evidence-backed problemPeer-reviewedHealth

The Challenges of Predicting Rare Outcomes: A Critical Appraisal of Machine Learning Using the Pediatric Resuscitation and Trauma Outcome (PRESTO) Model in a Tanzanian Injury Registry: These findings highlight the challenge of predicting a rare outcome, and emphasize the need to increase pediatric registry sample sizes to develop more accurate models for mortality risk stratification in LMICs.

Source article: The Challenges of Predicting Rare Outcomes: A Critical Appraisal of Machine Learning Using the Pediatric Resuscitation and Trauma Outcome (PRESTO) Model in a Tanzanian Injury Registry

World Journal of Surgery

Reader signal

How should this claim be treated?

Evidence-backed problemPeer-reviewedHealth

When they are addressed using artificial intelligence methods, though, average performance alone does not show whether a dataset is artificial intelligence ready, whether the endpoint is clinically valid, or whether errors are unevenly distributed across patient subgroups.

Source article: From Biomedical Datasets to Fairness-Aware Recommendations: An Integrated Data Orchestration Pipeline for Binary Clinical Predictions

Computational and Structural Biotechnology Journal

Reader signal

How should this claim be treated?

Evidence-backed problemPeer-reviewedHealth

Artificial intelligence-driven decision-making after endoscopic resection for early gastric cancer: Current guideline-based strategies, including the eCura system, provide structured risk stratification but rely on categorical decision-making and may lead to overtreatment, as nearly 90% of patients undergoing additional surgery do not have LNM.

Source article: Artificial intelligence-driven decision-making after endoscopic resection for early gastric cancer

Clinical Endoscopy

Reader signal

How should this claim be treated?

Evidence-backed problemPeer-reviewedLabor

Currently, there is a lack of comprehensive artificial intelligence frameworks capable of automating the entire meta-analysis workflow, including literature screening, data extraction, and quality assessment.

Source article: Unmasking bias in the evidence ecosystem: a panoramic analysis of 311,751 meta-analyses using an artificial intelligence agent-based approach

Journal of Global Health

Reader signal

How should this claim be treated?

Evidence-backed problemPeer-reviewedHealth

There is a growing literature on the prediction of risk of deterioration in hospital settings, including by leveraging artificial intelligence (AI) models.

Source article: A novel use of AI for prediction of clinical deterioration in a post-acute hospital

npj Health Systems

Reader signal

How should this claim be treated?

Evidence-backed problemPeer-reviewedScience

Explainable deep learning improves human mental models of self-driving cars: The opacity of these black-box planners makes it challenging to accurately anticipate when they will fail 4-6 , with potentially catastrophic consequences 7-9 .

Source article: Explainable deep learning improves human mental models of self-driving cars

Nature

Reader signal

How should this claim be treated?

Evidence-backed problemPeer-reviewedHealth

A novel computational solution, termed BioPINN-LM, integrates 2 computational methods for the rapid real-time prediction of aortic wall stress: a physics-informed neural network (PINN) trained with mechanical simulation data to predict wall stress and a multimodal large language model that uses output data from the PINN along with image-based geometry descriptors to provide interpretable risk assessments from both ends of the aorta.

Source article: Physics-Informed Neural Networks Meet Multimodal Large Language Models: Biomechanical Simulation in Aortic Aneurysm

Cyborg and Bionic Systems

Reader signal

How should this claim be treated?

Evidence-backed problemPeer-reviewedHealth

Four machine learning (ML) algorithms (Random Survival Forest, XGBoost, Elastic Net-regularized Cox, Support Vector Machine) were trained (80%) and tested (20%) to predict overall survival (OS).

Source article: Machine learning-integrated explainable artificial intelligence for survival prediction in urothelial carcinoma with enfortumab vedotin: an exploratory real-world analysis

Expert Opinion on Biological Therapy

Reader signal

How should this claim be treated?

Evidence-backed problemPeer-reviewedHealth

Objective To evaluate the accuracy, text quality, and perceived potential harm of EHR-integrated and LLM-generated hospital courses in pediatric inpatient care during early clinical implementation.

Source article: A Real-World Evaluation of Large Language Model-Generated Hospital Courses in Pediatrics

Hospital Pediatrics

Reader signal

How should this claim be treated?

Evidence-backed problemPeer-reviewedHealth

Artificial intelligence and ultra-high performance computing methods and experiments for drug discovery: virtual screening, deep learning, molecular dynamics simulations, ADMET modelling, and experimental validation: Furthermore, advancements in deep learning methodologies are required to improve the accuracy of prediction concerning target functionalities and ligand characteristics, even when complete receptor structures are not available.

Source article: Artificial intelligence and ultra-high performance computing methods and experiments for drug discovery: virtual screening, deep learning, molecular dynamics simulations, ADMET modelling, and experimental validation

Molecular Biomedicine