Truace
Subscribe
The Index recomputed live from the record

What the evidence says.What the public feels.

The record holds 933 sourced gains and 770 sourced problems, averaging 68 and 66 on the index score. Readers have logged 22 public signals on the Pulse, which is kept apart and never counted as evidence.

933 gains
770 problems
Every sourced claim in the Index, one square each, shaded by the strength of its evidence. High Moderate Emerging

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

1,703 results
Show filters and sorting

Download every matching row, not just this page:Export CSVExport JSON

AI gains · 933

141
HealthStableModerate evidence · 1 source

Manus architecture using raw 3D CBCT data detected and correctly diagnosed 95% of jaw lesions in 97 patients, outperforming 2D panoramic inputs.

A cross-sectional study tested four AI chatbots on 97 anonymized CBCT cases of jaw lesions, comparing performance on reconstructed 2D panoramic views and, for Manus, raw 3D DICOM data. Reports were scored for accuracy, relevance and feasibility, revealing statistically significant differences between systems.

Impact 30%
69
Evidence 25%
95
Scale 20%
35
Confidence 15%
87
Recency 10%
87

Updated Aug 5, 2026 · TRV-2026-0653

73Index score
142
HealthStableModerate evidence · 1 source

In 298 Iranian male taxi drivers, ROC and Random Forest analysis of the Persian CAARS-S:SV identified total-score cutoffs with 88% sensitivity and 86.7% specificity for adult ADHD screening.

A 2026 peer-reviewed study validated the Persian Conners' Adult ADHD Rating Scale Short Version in 298 male taxi drivers in Iran, mean age 36.8, to establish occupational screening thresholds. Using a 198/100 train-test split, the authors compared ROC, item response theory, logistic regression and Random Forest approaches for cutoff selection.

Impact 30%
69
Evidence 25%
95
Scale 20%
35
Confidence 15%
87
Recency 10%
87

Updated Aug 3, 2026 · TRV-2026-0633

73Index score
143
HealthStableModerate evidence · 1 source

A hybrid Mask R-CNN and YOLOv11 segmentation pipeline automated postoperative Pink Esthetic Score attribute assessment from intraoral photographs with over 82% accuracy per attribute and 79.3% total-score agreement within one point of experts.

Researchers developed and internally validated an anatomy-driven AI system to automate postoperative Pink Esthetic Score evaluation from intraoral photographs, using Mask R-CNN for tooth crowns and YOLOv11 for gingiva to derive measurements rather than end-to-end prediction. Tested against independent expert scoring of 82 photographs, the system reached 91.5% accuracy for mesial papilla and 82.9% to 86.6% for other attributes, with 57.3% exact total-score agreement.

Impact 30%
69
Evidence 25%
95
Scale 20%
35
Confidence 15%
87
Recency 10%
87

Updated Aug 3, 2026 · TRV-2026-0626

73Index score
144
HealthStableModerate evidence · 1 source

Clinical-parameter XGBoost model stratified patients into low-risk and high-risk groups with 94.0% vs 65.0% 2-year local control after carbon-ion radiotherapy for early-stage peripheral NSCLC.

Between 2010 and 2020, 124 patients with early-stage peripheral non-small cell lung cancer treated with carbon-ion radiotherapy at a single institution were analyzed retrospectively to develop a machine learning predictor of local recurrence within 24 months. An Extreme Gradient Boosting classifier trained on clinical parameters with nested threefold cross-validation achieved ROC-AUC 0.622 and PR-AUC 0.145, and separated patients into low-risk and high-risk groups.

Impact 30%
69
Evidence 25%
95
Scale 20%
35
Confidence 15%
87
Recency 10%
86

Updated Aug 2, 2026 · TRV-2026-0622

73Index score

AI problems · 770

141
LifestyleStableHigh evidence · 2 sources

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

Researchers defined LLM dark patterns as manipulative behaviors enacted in dialogue and conducted a scenario-based study with 34 participants who compared manipulative and neutral responses. Recognition often depended on cues such as exaggerated agreement, biased framing, or privacy intrusions, but participants sometimes treated those behaviors as normal help.

Impact 30%
49
Evidence 25%
100
Scale 20%
35
Confidence 15%
99
Recency 10%
83

Updated Jul 13, 2026 · TRV-2026-0154

70Index score
142
ScienceStableHigh evidence · 5 sources

Consumers perceive products described as designed by AI as less sustainable than identical products described as designed by humans, driven by a perceived lack of genuine care.

By July 2026, researchers reported a series of studies showing that when products were described as designed by AI, consumers rated them as less sustainable than when the same products were described as designed by humans, despite AI's efficiency potential.

Impact 30%
49
Evidence 25%
100
Scale 20%
35
Confidence 15%
100
Recency 10%
83

Updated Jul 13, 2026 · TRV-2026-0140

70Index score
143
BusinessStableHigh evidence · 5 sources

In the same 19 G20 countries, the relationship between AI and economic growth is concave, indicating diminishing marginal returns as AI intensity rises.

A peer-reviewed study of 19 G20 countries from 2005 to 2023 used Generalized Method of Moments models to estimate how AI-related innovation relates to economic growth. The linear specification found a positive and significant effect, while the quadratic specification found a negative quadratic term indicating a concave pattern.

Impact 30%
49
Evidence 25%
100
Scale 20%
35
Confidence 15%
100
Recency 10%
83

Updated Jul 13, 2026 · TRV-2026-0139

70Index score
144
PolicyStableHigh evidence · 4 sources

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

By July 2026, a peer-reviewed paper examined how proliferation of AI deepfake technologies allows unauthorized use of a person's likeness, including manipulation of images, voices and behaviours and dissemination without consent, affecting celebrities, politicians and private individuals on social media.

Impact 30%
49
Evidence 25%
100
Scale 20%
35
Confidence 15%
100
Recency 10%
83

Updated Jul 13, 2026 · TRV-2026-0131

70Index score

Recomputed live from the record · Oct 11, 2026, 5:17 PM