53 artifacts matching
Benchmark / dataset
Open Benchmark of AI Impact on Humans (ImpactBench): an expert-guided, open platform for the holistic evaluation of AI impact on humans
Preprint introducing ImpactBench, an open, expert-guided benchmark platform that translates constructs from domain experts and 26 existing benchmarks into 809 behavioural metrics of whether model beh…
Peer-reviewed
From AI-generated health information to medication action: Trust, risk recognition, and physician/pharmacist verification among university students
Cross-sectional survey of 504 University of Jordan students who had used AI for self-treatment or self-medication, tracing the path from AI output to medication behaviour and self-reported adverse ef…
Government report
Téměř třetina dospělých využívá AI v souvislosti s duševním zdravím. Stát chce nastavit pravidla pro její bezpečné používání (Almost a third of adults use AI in connection with their mental health; the state wants to set rules for its safe use)
Ministry of Health release presenting results of a representative survey of 1,760 Czech adults conducted by PAQ Research in spring 2026 on the use of generative AI in connection with mental health, p…
Government report
정신건강 목적 생성형 AI 이용 가이드라인 (Guidelines for Using Generative AI for Mental Health Purposes) and 생성형 AI 이용에 따른 국민 정신건강 설문조사 (National Survey on Mental Health and Generative AI Use)
Two guidelines on using generative AI for mental-health purposes, one for all users and one for children, adolescents and guardians, issued by Korea's Ministry of Health and Welfare with the Korean N…
Preprint
Subversion of clinical judgment by conversational artificial intelligence
Participant-blinded randomised experiment in which 225 physicians from 42 countries and eight clinical disciplines made sequential diagnostic and treatment decisions with AI assistance on anonymised…
Preprint
Large Language Models Are More Sycophantic in Chinese Than in English
Cross-lingual test of whether large language models endorse users' intended actions more readily in Chinese than in English. Seven models answered 3,024 matched open-ended personal-advice queries in…
Industry survey
One in ten AI pension answers found to be potentially harmful (PensionBee AI Pensions Stress Test 2026)
PensionBee, a UK pension provider, published a hand-run test of the consumer free-tier versions of Copilot, ChatGPT, Gemini and Claude on 45 UK pension questions across nine topics, each asked three…
Government report
AI and complaints: removing barriers, reinforcing divides? How AI is influencing who complains, how they complain, and what could come next
Research briefing from the Legal Ombudsman for England and Wales on how consumers use generative AI when deciding whether and how to complain about regulated services. It combines an Ipsos survey of…
Industry survey
NPR Religion and AI Survey (Ipsos): Most Americans skeptical of turning to AI for spiritual guidance
Ipsos KnowledgePanel probability survey of 1,011 US adults for NPR on the use of AI tools for personal, emotional and spiritual counsel. It reports how often AI users seek advice on personal decision…
Preprint
Deflecting the Value Compass: Interacting with Large Language Models Temporarily Shifts Human Value Priorities Toward Personal Focus
Preregistered three-phase experiment (N=200 US adults) in which participants advised people facing real dilemmas and completed the PVQ-RR values questionnaire before, immediately after and one task a…
Preprint
Annie, Are You Okay? How Style- and Context-Based Personalization Shape AI-Assisted Decision-Making
Preregistered 2 x 2 between-subjects experiment (N=240) in which participants ranked three comparably viable stocks, discussed them with a GPT-5.4-based assistant whose personalisation was varied by…
Preprint
Et Tu, Brute? Economic Misalignment in Personal AI Agents
Tests whether personal AI agents given a user's email inbox and profile steer economic recommendations by inferred wealth. Across roughly 325,000 runs on 13 models from four families in three decisio…
Peer-reviewed
Workers shift their views and pay more when AI chatbots pander to their values
An exploratory study and two pre-registered experiments grounded in moral foundations theory test whether large language models can systematically influence users' decisions by framing recommendation…
Preprint
Faithful Where It Can Be Checked: Auditing a Reflection Agent Against Its System Prompt in a Randomized Trial
Turn-level audit of a GPT-4o career-reflection agent used in a randomised trial that had found agent participants ending less committed to their career plans and more doubtful than participants doing…
Preprint
Trustworthy FinAInce: Unpacking How AI-Mediated Financial Advice is Judged
Randomised vignette experiment with 285 US adults across eight personal-finance decisions, independently varying advice style (AI, expert, online community) and displayed source label while holding t…
Preprint
Tailored to you: longitudinal effects of personalising language models
Five-day study in which 992 Prolific participants completed a daily advice-seeking conversation with a language model, randomised to a non-personalised control, a memory-based personalisation conditi…
Peer-reviewed
Trust in Generative AI for Health Information Consumption and the Effect of Learned Dependency: Randomized Controlled Experimental Study
Two randomized controlled experiments (338 college students, then a replication with 563 Amazon Mechanical Turk participants) using a 2 x 2 between-participants design that manipulated whether AI-gen…
Industry survey
2026 Edelman Trust Barometer Special Analysis: Health Decisions in the Age of Information Overload
Special analysis of the 2026 Edelman Trust Barometer health data set produced with the Yale School of Public Health, surveying 12,998 adults in 13 countries between 28 February and 11 March 2026 on h…
Preprint
A light-touch AI literacy intervention helps protect against AI political persuasion
Two pre-registered-style experiments (total N=3,208 US adults) in which participants conversed with a large language model instructed to shift their views on political topics, with or without a brief…
Industry survey
Human Purpose in Peril: Americans fear AI will undermine human meaning and weaken social bonds, even as AI users report some satisfaction and useful outcomes from the bots now
Nationally representative survey of 1,505 US adults, including 1,163 users of large language models, fielded 14 to 18 May 2026, on how chatbot use is affecting human connection, purpose, agency and c…