When ELIZA Meets Therapists: A Turing Test for the Heart and Mind DOI Open Access
Stephen Hatch, Zachary T. Goodman, Laura M. Vowels

и другие.

Опубликована: Май 8, 2024

“Can machines be therapists?” is a question receiving increased attention given the relative ease of working with generative artificial intelligence. Although recent (and decades-old) research has found that humans struggle to tell difference between responses from and humans1,2, findings suggest intelligence can write empathically generated content rated highly by therapists outperforms professionals.3,4,5,6,7,8 It uncertain if in preregistered competition where ChatGPT respond therapeutic vignettes about couple therapy whether a) panel participants (N = 830) which are ChatGPT-generated written therapists, b) or therapist-written fall more line key principles, c) linguistic differences conditions present. We show large sample rarely therapist, generally higher psychotherapy language patterns different. Using different measures, we then confirm were than therapist's suggesting these may explained part-of-speech response sentiment. This an early indication potential improve psychotherapeutic processes. anticipate this work lead development methods testing creating interventions. Further, discuss limitations (including lack context), how continued area improved efficacy interventions allowing such placed hands individuals who need them most.

Язык: Английский

Large Language Models in Mental Health Care: A Systematic Scoping Review (Preprint) DOI
Yining Hua, Fenglin Liu, Kailai Yang

и другие.

Опубликована: Июль 8, 2024

BACKGROUND The integration of large language models (LLMs) in mental health care is an emerging field. There a need to systematically review the application outcomes and delineate advantages limitations clinical settings. OBJECTIVE This aims provide comprehensive overview use LLMs care, assessing their efficacy, challenges, potential for future applications. METHODS A systematic search was conducted across multiple databases including PubMed, Web Science, Google Scholar, arXiv, medRxiv, PsyArXiv November 2023. All forms original research, peer-reviewed or not, published disseminated between October 1, 2019, December 2, 2023, are included without restrictions if they used developed after T5 directly addressed research questions RESULTS From initial pool 313 articles, 34 met inclusion criteria based on relevance LLM robustness reported outcomes. Diverse applications identified, diagnosis, therapy, patient engagement enhancement, etc. Key challenges include data availability reliability, nuanced handling states, effective evaluation methods. Despite successes accuracy accessibility improvement, gaps applicability ethical considerations were evident, pointing robust data, standardized evaluations, interdisciplinary collaboration. CONCLUSIONS hold substantial promise enhancing care. For full be realized, emphasis must placed developing datasets, development frameworks, guidelines, collaborations address current limitations.

Язык: Английский

Процитировано

14
Unge og helseinformasjon DOI Creative Commons
Marita Skjuve, Asbjørn Følstad, Kim Kristoffer Dysthe

и другие.

Tidsskrift for velferdsforskning, Год журнала: 2025, Номер 27(4), С. 1 - 17

Опубликована: Янв. 3, 2025

Mange unge etterspør informasjon om mental helse. Søkemotorer og sosiale medier er ofte innfallsporten til slik informasjon. Med lanseringen av ChatGPT i 2022 har flere tjenester basert på store språkmodeller, rettet seg mot unge, eksempelvis Snapchats My AI. Disse tjenestene brukervennlige gir umiddelbare svar spørsmål, men kan også generere feilaktig eller misvisende innhold. Det finnes lite kunnskap hvordan opplever spørsmål helse fra språkmodeller sammenlignet med fagpersoner. For å belyse dette gjennomførte vi en spørreundersøkelse der ba vurdere stor språkmodell (ChatGPT) fagpersoner ved informasjonstjeneste (ung.no). Utvalget besto 123 alderen 16–20 år. De fleste deltagerne var positive svarene ung.no ChatGPT. Samtidig anbefalte de oftere andre. Vi gjorde tillegg tematisk analyse åpne hvilket (ChatGPT fagpersoner) ville anbefale, hvorfor. Denne analysen ble gjort et mindre utvalg 24 deltagere for få mer forståelse kvantitative funnene. vektla informasjonskvalitet, opplevelse annerkjennelse velstrukturerte som begrunnelse sine preferanser.

Процитировано

0
AI in Relationship Counselling: Evaluating ChatGPT’s Therapeutic Capabilities in Providing Relationship Advice DOI Open Access
Laura M. Vowels, Rachel R. R. Francois‐Walcott, Joëlle Darwiche

и другие.

Опубликована: Окт. 29, 2023

Recent advancements in AI have led to chatbots, such as ChatGPT, capable of providing therapeutic responses. Early research evaluating chatbots’ ability provide relationship advice and single-session interventions has showed that both laypeople therapists rate them high on attributed empathy helpfulness. In the present study, 20 participants engaged intervention with ChatGPT were interviewed about their experiences. We evaluated performance comprising technical outcomes error linguistic accuracy quality questioning. The interviews analysed using reflexive thematic analysis which generated four themes: light at end tunnel; clearing fog; clinical skills; setting. analyses feasibility outcomes, coded by researchers perceived users, show provides realistic it consistently rated highly attributes skills, human-likeness, exploration, useability, clarity next steps for users’ problem. Limitations include a poor assessment risk reaching collaborative solutions participant. This study extends acceptance theories highlights potential capabilities support.

Язык: Английский

Процитировано

6
When ELIZA Meets Therapists: A Turing Test for the Heart and Mind DOI Open Access
Stephen Hatch, Zachary T. Goodman, Laura M. Vowels

и другие.

Опубликована: Май 8, 2024

“Can machines be therapists?” is a question receiving increased attention given the relative ease of working with generative artificial intelligence. Although recent (and decades-old) research has found that humans struggle to tell difference between responses from and humans1,2, findings suggest intelligence can write empathically generated content rated highly by therapists outperforms professionals.3,4,5,6,7,8 It uncertain if in preregistered competition where ChatGPT respond therapeutic vignettes about couple therapy whether a) panel participants (N = 830) which are ChatGPT-generated written therapists, b) or therapist-written fall more line key principles, c) linguistic differences conditions present. We show large sample rarely therapist, generally higher psychotherapy language patterns different. Using different measures, we then confirm were than therapist's suggesting these may explained part-of-speech response sentiment. This an early indication potential improve psychotherapeutic processes. anticipate this work lead development methods testing creating interventions. Further, discuss limitations (including lack context), how continued area improved efficacy interventions allowing such placed hands individuals who need them most.

Язык: Английский

Процитировано

2