TruaceTracing the truth around AITuesday, September 15, 2026
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.

1,419 results
Show filters and sorting

AI gains · 788

72
GainScience· Stable· Evidence: Moderate (1 source)

AI and machine learning techniques enable analysis of high-throughput scientific data to obtain insights, categorize, predict, and support evidence-based decisions across fundamental sciences.

On 2021-10-28 a peer-reviewed survey in The Innovation reviewed how artificial intelligence coupled with machine learning is being developed and applied across information science, mathematics, medical science, materials science, geoscience, life science, physics, and chemistry.

Impact 30%49
Evidence 25%95
Scale 20%60
Confidence 15%87
Recency 10%88

Updated Jul 13, 2026 · TRV-2026-0207

72
GainSports· Stable· Evidence: Moderate (1 source)

Machine learning models demonstrated promising accuracy for sport applications including action recognition, injury prediction and prevention, and athlete selection, offering opportunities for performance enhancement and decision-making.

A scoping review published May 25, 2026 examined 270 peer-reviewed studies from 2002 to 2024 on machine learning in sport. It found applications across 12 subject areas, most frequently computer science, biomechanics, and sport psychology, with common uses in action recognition, injury prediction/prevention, and athlete selection/talent identification.

Impact 30%49
Evidence 25%95
Scale 20%60
Confidence 15%87
Recency 10%88

Updated Jul 13, 2026 · TRV-2026-0186

72
GainCrime· Stable· Evidence: Moderate (1 source)

Listeners detected ElevenLabs-cloned speech more accurately when audio was transmitted through narrowband GSM (AMR-NB) at 63.7%, compared to other telecom conditions.

By June 2026, researchers tested how telecom transmission affects human detection of cloned speech. They created natural and ElevenLabs-synthesized utterances from nine speakers, processed them through simulated GSM, VoLTE, and VoIP codecs, and asked 95 participants to classify them as human or synthetic.

Impact 30%63
Evidence 25%95
Scale 20%35
Confidence 15%87
Recency 10%88

Updated Jul 13, 2026 · TRV-2026-0152

71
GainHealth· Stable· Evidence: High (2 sources)

AI-driven wearable bioelectronics enable continuous monitoring of cardiac activity, glucose and biomarkers to support early disease detection, chronic disease management and remote patient monitoring.

As of the June 2025 review, integration of AI with wearable bioelectronics was presented as enabling proactive, personalized monitoring of cardiac activity, glucose levels and biomarkers, with applications in early detection, chronic condition management and precision therapeutics.

Impact 30%49
Evidence 25%100
Scale 20%35
Confidence 15%99
Recency 10%90

Updated Jul 23, 2026 · TRV-2026-0520

AI problems · 631

68
ProblemHealth· Stable· Evidence: Moderate (1 source)

In the same vignettes, experts identified safety risks in AI-generated discharge instructions not found in nurse texts, and nurses significantly outperformed AI on empathy and readability.

A prospective double-blind vignette study compared discharge instructions created by GPT-4 and by five registered nurses across five standardized scenarios. Fifteen experts rated accuracy and 38 patients rated empathy and readability, with NLP analysis of text complexity, using paired tests and a generalized linear mixed model.

Impact 30%49
Evidence 25%95
Scale 20%35
Confidence 15%87
Recency 10%97

Updated Sep 1, 2026 · TRV-2026-0952

68
ProblemHealth· Stable· Evidence: Moderate (1 source)

In DLBCL patients treated with first-line immunochemotherapy, CT-measured sarcopenia in the lowest tertile of muscle mass is associated with inferior overall survival driven by nonrelapse mortality and higher risk of hematologic toxicity.

In patients with newly diagnosed diffuse large B-cell lymphoma from the PETAL trial, investigators used machine learning-supported body composition analysis of CT imaging to measure skeletal muscle mass. Those in the lowest tertile had inferior survival after adjustment for established risk factors, with cause-specific analyses pointing to nonrelapse mortality rather than lymphoma-specific death.

Impact 30%49
Evidence 25%95
Scale 20%35
Confidence 15%87
Recency 10%97

Updated Sep 1, 2026 · TRV-2026-0948

68
ProblemHealth· Stable· Evidence: Moderate (1 source)

AI simulations failed to accurately reproduce quantitative facial anthropometric changes after denture placement despite visual similarity.

A prospective study of 14 edentulous patients compared actual post-denture facial photographs with AI-generated predictions from pretreatment images using Gemini and FaceApp, assessing patient preference, expert esthetic ratings, and quantitative anthropometric measurements.

Impact 30%49
Evidence 25%95
Scale 20%35
Confidence 15%87
Recency 10%97

Updated Aug 31, 2026 · TRV-2026-0938

68
ProblemClimate· Stable· Evidence: Moderate (1 source)

Many ML studies of soil potentially toxic elements rely on spatially naive validation, and random cross-validation often overestimates predictive performance when spatial dependence is ignored, with incomplete uncertainty reporting.

Published August 29, 2026, this peer-reviewed synthesis reviews 2020-2025 literature on machine learning for mapping potentially toxic elements in soils. It finds growing use of ML with environmental covariates but persistent use of spatially naive validation, limited interpretation, and incomplete uncertainty reporting.

Impact 30%49
Evidence 25%95
Scale 20%35
Confidence 15%87
Recency 10%97

Updated Aug 31, 2026 · TRV-2026-0932

Recomputed live from the record · Sep 16, 2026, 12:16 AM