For people who want the doors
Cited field
Public numbers about how AI is used, how it is regulated, and how often it flatters. Each line is a source you can open. None of this is a TexanoAI outcome claim.
Fact = the source said it. Assumption = we think it matters to the field. Projection forbidden: typical-user ROI, “we caused this,” clinical cure.
Work and adoption
- About half of employed U.S. adults use AI in their role at least a few times a year. Gallup workplace AI, 2026
- U.S. employee engagement has stayed flat near 31% while AI adoption rises. Gallup engagement + AI
- Organizations reporting AI use: 55% (2023) → 78% (2024). Generative AI in at least one business function: 33% → 71%. Stanford HAI AI Index 2025 · Economy
- U.S. private AI investment $109.1 billion in 2024 — nearly 12× China, 24× the U.K. in that Index cut. Stanford HAI · Economy
- AI-related incidents logged at 233 in 2024 (+56.4% vs 2023). Stanford HAI · 10 charts
Law and the patchwork
- U.S. state AI-related laws: 1 (2016) → 49 (2023) → 131 in the following year. Stanford HAI AI Index 2025 · Policy · our note
- U.S. federal AI-related regulations: 25 (2023) → 59 (2024), from 42 agencies. Same Policy chapter
- EU AI Act — official Commission page. European Commission · our note
- Japan METI + MIC: AI Guidelines for Business Ver 1.0 (English press). 19 April 2024
- Guidelines Ver 1.0 English PDF (provisional). PDF
- Guidelines Ver 1.2 English PDF (31 March 2026) — includes the 2025 AI Promotion Act. PDF · eight principles on this site · Japan AIA essay
Honesty under load (appeasement)
- OpenAI rolled back a GPT-4o update in April 2025 for being “too sycophant-y.” OpenAI
- Models affirmed user actions about 47–50% more often than humans. Cheng et al., arXiv:2510.01395
- SycEval: overall sycophancy near 58%. Fanous et al.
- Gartner: more than 40% of agentic AI projects expected cancelled by end of 2027. The Register / Gartner
Companions and substitution (context, not a sales hook)
- U.S. Surgeon General, 2023: loneliness and isolation advisory. HHS PDF
- De Freitas et al.: short-term companion interaction and loneliness scales. arXiv:2407.19096
- Maples, Cerit, Vishwanath, Pea — Replika student survey. npj
- Zhang et al. on Character.AI: intensive companionship use and lower well-being. arXiv:2506.12605
- OpenAI–MIT affective-use study. OpenAI
- Common Sense Media: most U.S. teens have tried a companion chatbot. Talk, Trust, and Trade-Offs
- Essay that holds this together: The Weird Science Dilemma
Educational. Not legal advice. Not a clinic. Check the original page before any compliance or health decision.