+ HEALTH Background Discharge letters are essential for continuity of care, yet patients often misunderstand their con…± HEALTH Hematoxylin and Eosin (H&E) glass slide preparation is fundamental to histopathology. Pathologists' expertise…+ CLIMATE Traditional soil heavy metal risk assessments suffer from fragmented indices and poor logical integration. Co…± HEALTH Background Neonatal pneumothorax can progress rapidly, yet detection on supine chest radiographs remains chal…+ HEALTH T cell exhaustion has widespread implications for the progression and treatment of chronic diseases including…+ HEALTH Background Standard pulmonary arterial hypertension (PAH) risk models may not fully apply to PAH associated w…+ HEALTH Objectives To map and synthesize the scientific evidence regarding the use of Augmented Reality (AR) and Mixe…± HEALTH Objective To examine whether the outputs of machine learning algorithms designed to predict risk of cardiovas…
TruaceTracing the truth around AIMonday, September 14, 2026
Subscribe
The Index

What the evidence says.What the public feels.

Ranks distinct AI gain and problem claims from the published record. Scores reward impact, independent source strength, scale, confidence, and recency.

1,402 results
Show filters and sorting

AI gains · 778

67
GainScience· Stable· Evidence: Moderate (1 source)

The AI Scientist pipeline autonomously executed the entire research lifecycle and produced a manuscript that cleared initial peer review at a selective machine learning workshop.

Researchers built The AI Scientist, an agentic system using foundation models to automate conception, coding, experimentation, data analysis, manuscript writing, and peer review. By March 2026 they reported that a manuscript fully generated by the system passed first-round review for a workshop at a top-tier machine learning conference.

Impact 30%49
Evidence 25%95
Scale 20%35
Confidence 15%87
Recency 10%88

Updated Jul 13, 2026 · TRV-2026-0163

67
GainEducation· Stable· Evidence: Moderate (1 source)

Students reported significantly lower anxiety during AIvaluate-mediated performance-based assessments compared to traditional face-to-face assessments, with usability rated in the good range.

Researchers compared traditional face-to-face performance-based assessments with sessions mediated by AIvaluate, an LLM-augmented emotionally intelligent conversational agent, using 35 pre-university students in a counterbalanced within-subjects design measuring emotional state, usability and qualitative perceptions.

Impact 30%49
Evidence 25%95
Scale 20%35
Confidence 15%87
Recency 10%88

Updated Jul 13, 2026 · TRV-2026-0162

67
GainEducation· Stable· Evidence: Moderate (1 source)

AI use was associated with strong perceived gains in educational support through enhanced and personalized learning environments, and in emotional and mental well-being.

On 2026-06-04 a peer-reviewed study reported results from 150 participants surveyed about AI's influence on behavior and well-being. Using a Likert-scale questionnaire analyzed in SPSS version 25, the authors found educational support had the highest mean at 4.89 and emotional and mental well-being had the highest mean difference at 4.9, with all T-tests significant at p < 0.000.

Impact 30%49
Evidence 25%95
Scale 20%35
Confidence 15%87
Recency 10%88

Updated Jul 13, 2026 · TRV-2026-0161

67
GainLabor· Stable· Evidence: Moderate (1 source)

Passive AI use initially boosted enjoyment and satisfaction, while active collaboration where workers drafted first then used AI to refine preserved psychological connection comparable to independent work.

Researchers ran a pre-registered lab experiment with 269 participants doing occupation-specific writing under no AI, passive AI copying, or active drafting-then-refining, plus a 270-person real-world survey. Passive copying reduced self-efficacy, ownership, and meaningfulness, with efficacy and meaningfulness losses persisting after returning to manual work, while active collaboration preserved connection similar to working alone.

Impact 30%49
Evidence 25%95
Scale 20%35
Confidence 15%87
Recency 10%88

Updated Jul 13, 2026 · TRV-2026-0160

67
GainEducation· Stable· Evidence: Moderate (1 source)

Chinese high school EFL students showed a significant time effect and three-phase trajectory in writing complexity, accuracy and fluency during GenAI-assisted continuation task instruction.

In a 9-week repeated-measures study, 25 Chinese high school EFL students completed five continuation writing tasks at two-week intervals while receiving GenAI-assisted instruction, producing 125 samples analyzed with generalized estimating equation analysis under Dynamic Systems Theory.

Impact 30%49
Evidence 25%95
Scale 20%35
Confidence 15%87
Recency 10%88

Updated Jul 13, 2026 · TRV-2026-0159

67
GainPolicy· Stable· Evidence: Moderate (1 source)

Authors derive a two-tier structure of auditable design constraints that operationalizes ethical principles for AI-mediated afterlife systems, providing a concrete bridge to governance.

As of the June 2026 publication date, the authors describe a rapid proliferation of AI-mediated digital afterlife technologies and a growing ethical literature on their risks, without a matching operational framework. They propose a nine-dimensional taxonomy and a two-tier constraint model where consent, fidelity/disclosure, and purpose serve as threshold conditions for permissibility.

Impact 30%49
Evidence 25%95
Scale 20%35
Confidence 15%87
Recency 10%88

Updated Jul 13, 2026 · TRV-2026-0153

67
GainCrime· Stable· Evidence: Moderate (1 source)

In a preregistered classification task, typical adults distinguished real from AI-generated faces and real from AI-generated voices at rates above chance.

By June 2026, researchers had tested whether people can tell real from synthetic faces and voices and whether that skill transfers across senses. In a preregistered study, participants classified both types of stimuli, and performance for each modality was significantly above chance when measured with signal detection theory.

Impact 30%49
Evidence 25%95
Scale 20%35
Confidence 15%87
Recency 10%88

Updated Jul 13, 2026 · TRV-2026-0151

67
GainEducation· Stable· Evidence: Moderate (1 source)

Preservice teachers reported that LLMs helped during early modeling work by supporting assumption-making, formula retrieval, and building models and solution strategies.

By April 2026, researchers studied 150 mastere28099s-level preservice teachers at a German university as they collaboratively solved three authentic mathematical modeling problems with large language models, analyzing interaction worksheets, surveys, and interviews.

Impact 30%49
Evidence 25%95
Scale 20%35
Confidence 15%87
Recency 10%88

Updated Jul 13, 2026 · TRV-2026-0148

AI problems · 624

No problems on this page.

Recomputed live from the record · Sep 14, 2026, 1:10 PM