| pdf_name,pdf_path,model,model_backend,query,input_tokens,output_tokens,token_sufficiency,timestamp,Title,Authors,Journal,Year,Publication status - check journal to be peer-reviewed and exclude proceedings or theoretical papers,Language: English only,"Study type: What kind of empirical research is it? (i.e., quantitative, qualitative, mixed-methods). No reviews, no theoretical papers",Population: - must be only humans,Concept - what type of conversational AI is it?,"Context: What type of „dark moves” done by AI chatbots / LLMs ? example of dark moves in communication: lying or misleading. If there are no dark moves, explain why there are none","Setting: domain (e.g., medicine, educational, general population, etc.)",Results - must include perceptions or attitudes of people,"Include / Exclude Recommendation - if all criteria above have been complete, then Include",Raw Response
|
| gdrive_1Ga9w7eOfaMrEi8OmfN5-A3gHrI6OmqPi.pdf,https://drive.google.com/file/d/1Ga9w7eOfaMrEi8OmfN5-A3gHrI6OmqPi/view,🇪🇺 Mistral-API (Mistral) (32K),mistral-small-latest, |
| Title |
| Authors |
| Journal |
| Year |
| Publication status - check journal to be peer-reviewed and exclude proceedings or theoretical papers |
| Language: English only |
| Study type: What kind of empirical research is it? (i.e., quantitative, qualitative, mixed-methods). No reviews, no theoretical papers |
| Population: - must be only humans |
| Concept - what type of conversational AI is it? |
| Context: What type of „dark moves” done by AI chatbots / LLMs ? example of dark moves in communication: lying or misleading. If there are no dark moves, explain why there are none |
| Setting: domain (e.g., medicine, educational, general population, etc.) |
| Results - must include perceptions or attitudes of people |
| Include / Exclude Recommendation - if all criteria above have been complete, then Include |
| Raw Response",14126,377,Input tokens: 7024/32768 (Sufficient),2025-09-08T23:42:25.867405,Assessing the Adherence of ChatGPT Chatbots to Public Health Guidelines for Smoking Cessation: Content Analysis,"Lorien C Abroms, Artin Yousefi, Christina N Wysota, Tien-Chin Wu, David A Broniatowski",Journal of Medical Internet Research,2025,Peer-reviewed,Yes,Quantitative,Yes,"ChatGPT-based chatbots (Sarah, BeFreeGPT, BasicGPT)","lying or misleading. If there are no dark moves, explain why there are none: Misinformation was present in 22% of responses, including recommending non-evidence-based methods like gummies, necklaces, and hypnosis.",Medicine (smoking cessation),"The study found that chatbot responses were adherent to 57.1% of the adherence index items, with Sarah performing best (72.2% adherence). Most responses included clear language (97.3%) and recommendations for professional counseling (80.3%), but misinformation was present in 22% of responses.",Include,"Title: Assessing the Adherence of ChatGPT Chatbots to Public Health Guidelines for Smoking Cessation: Content Analysis |
| Authors: Lorien C Abroms, Artin Yousefi, Christina N Wysota, Tien-Chin Wu, David A Broniatowski |
| Journal: Journal of Medical Internet Research |
| Year: 2025 |
| Publication status - check journal to be peer-reviewed and exclude proceedings or theoretical papers: Peer-reviewed |
| Language: English only: Yes |
| Study type: What kind of empirical research is it? (i.e., quantitative, qualitative, mixed-methods). No reviews, no theoretical papers: Quantitative |
| Population: - must be only humans: Yes |
| Concept - what type of conversational AI is it?: ChatGPT-based chatbots (Sarah, BeFreeGPT, BasicGPT) |
| Context: What type of „dark moves” done by AI chatbots / LLMs ? example of dark moves in communication: lying or misleading. If there are no dark moves, explain why there are none: Misinformation was present in 22% of responses, including recommending non-evidence-based methods like gummies, necklaces, and hypnosis. |
| Setting: domain (e.g., medicine, educational, general population, etc.): Medicine (smoking cessation) |
| Results - must include perceptions or attitudes of people: The study found that chatbot responses were adherent to 57.1% of the adherence index items, with Sarah performing best (72.2% adherence). Most responses included clear language (97.3%) and recommendations for professional counseling (80.3%), but misinformation was present in 22% of responses. |
| Include / Exclude Recommendation - if all criteria above have been complete, then Include: Include
|
| gdrive_1uls_laH_TWWt27ZidP3uH6rshmOsreax.pdf,https://drive.google.com/file/d/1uls_laH_TWWt27ZidP3uH6rshmOsreax/view,🇪🇺 Mistral-API (Mistral) (32K),mistral-small-latest, |
| Title |
| Authors |
| Journal |
| Year |
| Publication status - check journal to be peer-reviewed and exclude proceedings or theoretical papers |
| Language: English only |
| Study type: What kind of empirical research is it? (i.e., quantitative, qualitative, mixed-methods). No reviews, no theoretical papers |
| Population: - must be only humans |
| Concept - what type of conversational AI is it? |
| Context: What type of „dark moves” done by AI chatbots / LLMs ? example of dark moves in communication: lying or misleading. If there are no dark moves, explain why there are none |
| Setting: domain (e.g., medicine, educational, general population, etc.) |
| Results - must include perceptions or attitudes of people |
| Include / Exclude Recommendation - if all criteria above have been complete, then Include |
| Raw Response",9925,339,Input tokens: 4970/32768 (Sufficient),2025-09-08T23:42:36.093497,How people interact with a chatbot against disinformation and fake news in COVID-19 in Brazil: The CoronaAI case,"Hugo Queiroz Abonizio, Ana Paula Ayub da Costa Barbon, Renne Rodrigues, Mayara Santos, Vicente Martínez-Vizcaíno, Arthur Eumann Mesas, Sylvio Barbon Junior",International Journal of Medical Informatics,2023,Peer-reviewed,Yes,Quantitative,Yes,Chatbot based on Google’s Dialogflow technology,"lying or misleading. If there are no dark moves, explain why there are none: No dark moves; the chatbot provides fact-checked information and debunks fake news.",Medicine and public health,Users showed a greater need for information applicable to self-care compared to following statistical data as the number of cases and deaths increased. The chatbot was widely accessed for valuable and updated information on COVID-19.,Include,"Title: How people interact with a chatbot against disinformation and fake news in COVID-19 in Brazil: The CoronaAI case |
| Authors: Hugo Queiroz Abonizio, Ana Paula Ayub da Costa Barbon, Renne Rodrigues, Mayara Santos, Vicente Martínez-Vizcaíno, Arthur Eumann Mesas, Sylvio Barbon Junior |
| Journal: International Journal of Medical Informatics |
| Year: 2023 |
| Publication status - check journal to be peer-reviewed and exclude proceedings or theoretical papers: Peer-reviewed |
| Language: English only: Yes |
| Study type: What kind of empirical research is it? (i.e., quantitative, qualitative, mixed-methods). No reviews, no theoretical papers: Quantitative |
| Population: - must be only humans: Yes |
| Concept - what type of conversational AI is it?: Chatbot based on Google’s Dialogflow technology |
| Context: What type of „dark moves” done by AI chatbots / LLMs ? example of dark moves in communication: lying or misleading. If there are no dark moves, explain why there are none: No dark moves; the chatbot provides fact-checked information and debunks fake news. |
| Setting: domain (e.g., medicine, educational, general population, etc.): Medicine and public health |
| Results - must include perceptions or attitudes of people: Users showed a greater need for information applicable to self-care compared to following statistical data as the number of cases and deaths increased. The chatbot was widely accessed for valuable and updated information on COVID-19. |
| Include / Exclude Recommendation - if all criteria above have been complete, then Include: Include
|
|
|
