Mannan Kazi Abdul
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Mannan Kazi Abdul has a total of 585 co-authors. Here are the top co-authors by number of shared publications:
- Farhana Khandaker Mursheda
Shanto-Mariam University of Creative Technology
22 shared publications - Joya Joyshree Das
Supreme Court of Bangladesh
11 shared publications - Akter Sraboni
Shanto-Mariam University of Creative Technology, Bangladesh
8 shared publications Khandaker Mursheda Farhana
8 shared publications- Chowdhury Afrina Afroze
Shanto-Mariam University of Creative Technology
6 shared publications - Al Abdullah
Shanto-Mariam University of Creative Technology
5 shared publications
Latest Publications
- Open Access
- 19 November 2026
- Published by Springer Nature in Research Square
Caution about AI Generated News Editorial Accountability and Support for Proportionate Regulation in Bangladesh
Generative artificial intelligence complicates the relationship between news credibility and demands for regulation. Caution about substantially AI-generated news need not imply rejection of editorially supervised assistance, while support for accountability need not imply support for unrestricted content controls. This study examines these distinctions in 385 adult, AI-aware questionnaire records supplied as empirical data from Bangladesh. It analyses five separate statements concerning caution, labelling, accuracy accountability, oversight-related trust and proportionate regulation, with supporting awareness and verification items. Low five-item block reliabilities preclude treating the questionnaire as a validated latent-variable model. Analyses combine response distributions, paired contrasts, joint endorsement patterns, rank correlations and exploratory adjusted logistic regression. Accuracy accountability received 84.2% endorsement and oversight-related trust 83.1%, compared with 69.4% for caution, 68.8% for appropriate labelling and 66.8% for proportionate rules respecting lawful expression. Caution and oversight-related trust were jointly endorsed by 57.7%. Accountability exceeded rules endorsement by 17.4 percentage points, while similar aggregate labelling and rules percentages concealed differing responses in 170 records. No selected correlation or focal binary-model coefficient met the 5% threshold after Holm adjustment. The oversight coefficient was positive but sensitive to response coding and sample restriction. The findings distinguish epistemic caution, editorial responsibility and conditional regulatory support rather than demonstrating a causal trust-to-regulation pathway. They support evaluating particular disclosures and oversight procedures against both informational outcomes and expressive rights. Mixed respondent status, undocumented recruitment and fieldwork dates, single-item measurement and unresolved source-documentation labels constrain generalisation. The contribution is an item-specific account of news governance preferences that does not equate favourable survey responses with verified competence or a mandate for coercive regulation.
- 19 November 2026
- Published by Springer Nature in Research Square
Open AccessCaution about AI Generated News Editorial Accountability and Support for Proportionate Regulation in Bangladesh
Generative artificial intelligence complicates the relationship between news credibility and demands for regulation. Caution about substantially AI-generated news need not imply rejection of editorially supervised assistance, while support for accountability need not imply support for unrestricted content controls. This study examines these distinctions in 385 adult, AI-aware questionnaire records supplied as empirical data from Bangladesh. It analyses five separate statements concerning caution, labelling, accuracy accountability, oversight-related trust and proportionate regulation, with supporting awareness and verification items. Low five-item block reliabilities preclude treating the questionnaire as a validated latent-variable model. Analyses combine response distributions, paired contrasts, joint endorsement patterns, rank correlations and exploratory adjusted logistic regression. Accuracy accountability received 84.2% endorsement and oversight-related trust 83.1%, compared with 69.4% for caution, 68.8% for appropriate labelling and 66.8% for proportionate rules respecting lawful expression. Caution and oversight-related trust were jointly endorsed by 57.7%. Accountability exceeded rules endorsement by 17.4 percentage points, while similar aggregate labelling and rules percentages concealed differing responses in 170 records. No selected correlation or focal binary-model coefficient met the 5% threshold after Holm adjustment. The oversight coefficient was positive but sensitive to response coding and sample restriction. The findings distinguish epistemic caution, editorial responsibility and conditional regulatory support rather than demonstrating a causal trust-to-regulation pathway. They support evaluating particular disclosures and oversight procedures against both informational outcomes and expressive rights. Mixed respondent status, undocumented recruitment and fieldwork dates, single-item measurement and unresolved source-documentation labels constrain generalisation. The contribution is an item-specific account of news governance preferences that does not equate favourable survey responses with verified competence or a mandate for coercive regulation.
- Open Access
- 18 November 2026
- Published by Elsevier in SSRN Electronic Journal
Deepfake Awareness and Self Reported Verification among AI aware Respondents in Bangladesh
Deepfake education is often expected to encourage verification, yet awareness, beliefs about intention and actual checking require different evidence. This study examines these distinctions in 385 adult, AI-aware survey records from Bangladesh, including a descriptive subset of 227 law students. The questionnaire separately asks about awareness of convincing AI-generated media, source checking, corroboration, the perceived influence of awareness on verification intention, and intended restraint in sharing unverified material. Responses were analysed using item distributions, joint endorsement, paired comparisons, exploratory rank correlations and threshold sensitivity summaries. Overall, 244 respondents (63.4%) endorsed deepfake awareness, 265 (68.8%) reported source checking, and 278 (72.2%) reported seeking corroboration. The statement that awareness increases verification intention received 327 endorsements (84.9%), exceeding source-checking endorsement by 16.1 percentage points (pointwise bootstrap 95% interval 10.1-21.8). However, awareness had small ordinal correlations with checking and corroboration (ρ = 0.084 and 0.077); neither was statistically distinguishable from zero after adjustment within the four-test family. Law-student summaries showed a similar ordering of endorsement. These findings concern differently worded self-reports and do not establish an intention-behaviour gap, intervention effect or inability to identify deepfakes. The article contributes a measurement-based distinction between endorsing an awareness-to-intention proposition and independently observing verification. It recommends evaluating media-literacy education through source tracing, corroboration and discernment tasks that include authentic as well as manipulated content. Undocumented recruitment and the mixed sample limit population inference.
- 18 November 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessDeepfake Awareness and Self Reported Verification among AI aware Respondents in Bangladesh
Deepfake education is often expected to encourage verification, yet awareness, beliefs about intention and actual checking require different evidence. This study examines these distinctions in 385 adult, AI-aware survey records from Bangladesh, including a descriptive subset of 227 law students. The questionnaire separately asks about awareness of convincing AI-generated media, source checking, corroboration, the perceived influence of awareness on verification intention, and intended restraint in sharing unverified material. Responses were analysed using item distributions, joint endorsement, paired comparisons, exploratory rank correlations and threshold sensitivity summaries. Overall, 244 respondents (63.4%) endorsed deepfake awareness, 265 (68.8%) reported source checking, and 278 (72.2%) reported seeking corroboration. The statement that awareness increases verification intention received 327 endorsements (84.9%), exceeding source-checking endorsement by 16.1 percentage points (pointwise bootstrap 95% interval 10.1-21.8). However, awareness had small ordinal correlations with checking and corroboration (ρ = 0.084 and 0.077); neither was statistically distinguishable from zero after adjustment within the four-test family. Law-student summaries showed a similar ordering of endorsement. These findings concern differently worded self-reports and do not establish an intention-behaviour gap, intervention effect or inability to identify deepfakes. The article contributes a measurement-based distinction between endorsing an awareness-to-intention proposition and independently observing verification. It recommends evaluating media-literacy education through source tracing, corroboration and discernment tasks that include authentic as well as manipulated content. Undocumented recruitment and the mixed sample limit population inference.
- Open Access
- 18 November 2026
- Published by Springer Nature in Research Square
Generative AI Copyright Awareness and Views on Human Creative Contribution in Bangladesh
Generative artificial intelligence (AI) raises questions about human creative contribution, copyright in outputs and third-party rights. This study examines responses to copyright-related statements in 385 adult, AI-aware survey records from Bangladesh, including a separately described subgroup of 241 law students. The questionnaire asks about recognition of copyright questions, the relevance of human contribution, terms of use, third-party rights and the perceived value of guidance. It does not ask respondents to assign ownership to a specified actor or solve a legal problem; consequently, it does not measure actual copyright knowledge or establish legal ownership. Items were analysed separately using descriptive distributions, paired endorsement comparisons and exploratory rank correlations. In the full sample, 191 respondents (49.6%) endorsed recognising that copyright questions can arise, 223 (57.9%) endorsed the relevance of human contribution, and 303 (78.7%) agreed that clear guidance would improve their responsible use. Among law students, the respective percentages were 51.0%, 60.6% and 80.5%. Overall, 147 respondents endorsed both recognising copyright questions and the potential usefulness of guidance, while 156 endorsed guidance without endorsing recognition. This pattern is not evidence that guidance improves knowledge or behaviour. The article distinguishes statements of awareness and normative expectations from demonstrated legal competence, and recommends scenario-based assessment and documentation of respondent recruitment for future research.
- 18 November 2026
- Published by Springer Nature in Research Square
Open AccessGenerative AI Copyright Awareness and Views on Human Creative Contribution in Bangladesh
Generative artificial intelligence (AI) raises questions about human creative contribution, copyright in outputs and third-party rights. This study examines responses to copyright-related statements in 385 adult, AI-aware survey records from Bangladesh, including a separately described subgroup of 241 law students. The questionnaire asks about recognition of copyright questions, the relevance of human contribution, terms of use, third-party rights and the perceived value of guidance. It does not ask respondents to assign ownership to a specified actor or solve a legal problem; consequently, it does not measure actual copyright knowledge or establish legal ownership. Items were analysed separately using descriptive distributions, paired endorsement comparisons and exploratory rank correlations. In the full sample, 191 respondents (49.6%) endorsed recognising that copyright questions can arise, 223 (57.9%) endorsed the relevance of human contribution, and 303 (78.7%) agreed that clear guidance would improve their responsible use. Among law students, the respective percentages were 51.0%, 60.6% and 80.5%. Overall, 147 respondents endorsed both recognising copyright questions and the potential usefulness of guidance, while 156 endorsed guidance without endorsing recognition. This pattern is not evidence that guidance improves knowledge or behaviour. The article distinguishes statements of awareness and normative expectations from demonstrated legal competence, and recommends scenario-based assessment and documentation of respondent recruitment for future research.
- Open Access
- 17 November 2026
- Published by Springer Nature in Research Square
Legal Awareness and Responsible Responses to AI Generated Misinformation in Bangladesh
Generative artificial intelligence can produce convincing misleading text and synthetic media, raising questions about how people assess and respond to online information. This study examines self-reported awareness of possible legal consequences and three distinct responses to potentially false content: considering harm before forwarding, reporting deceptive material and correcting or removing content later found false. It analyses 385 adult, AI-aware questionnaire records from a Bangladesh sample comprising law students, graduates, legal professionals and other respondents. Because the questionnaire does not measure the frequency of exposure to AI misinformation or observed social media activity, the study assesses stated awareness and self-reported practices rather than exposure or actual conduct. Items are analysed separately using descriptive distributions, paired comparisons and exploratory Spearman correlations with Holm adjustment. Legal-consequence awareness was endorsed by 244 respondents (63.4%); consideration of harm by 269 (69.9%); reporting by 268 (69.6%); and correction or removal by 259 (67.3%). Awareness and reporting showed a small positive rank correlation (ρ = .138), whereas the relationships with harm consideration and correction were uncertain after adjustment for three tests. Differences between correlation coefficients require direct comparison; statistical significance in one association alone does not demonstrate a difference across practices. The findings do not show that legal awareness causes responsible behaviour. The article integrates information credibility, media literacy and the theory of planned behaviour as perspectives for a future study with measured exposure, validated items and observed verification tasks.
- 17 November 2026
- Published by Springer Nature in Research Square
Open AccessLegal Awareness and Responsible Responses to AI Generated Misinformation in Bangladesh
Generative artificial intelligence can produce convincing misleading text and synthetic media, raising questions about how people assess and respond to online information. This study examines self-reported awareness of possible legal consequences and three distinct responses to potentially false content: considering harm before forwarding, reporting deceptive material and correcting or removing content later found false. It analyses 385 adult, AI-aware questionnaire records from a Bangladesh sample comprising law students, graduates, legal professionals and other respondents. Because the questionnaire does not measure the frequency of exposure to AI misinformation or observed social media activity, the study assesses stated awareness and self-reported practices rather than exposure or actual conduct. Items are analysed separately using descriptive distributions, paired comparisons and exploratory Spearman correlations with Holm adjustment. Legal-consequence awareness was endorsed by 244 respondents (63.4%); consideration of harm by 269 (69.9%); reporting by 268 (69.6%); and correction or removal by 259 (67.3%). Awareness and reporting showed a small positive rank correlation (ρ = .138), whereas the relationships with harm consideration and correction were uncertain after adjustment for three tests. Differences between correlation coefficients require direct comparison; statistical significance in one association alone does not demonstrate a difference across practices. The findings do not show that legal awareness causes responsible behaviour. The article integrates information credibility, media literacy and the theory of planned behaviour as perspectives for a future study with measured exposure, validated items and observed verification tasks.
- Open Access
- 16 November 2026
- Published by Springer Nature in Research Square
Farmers Rights Information and Safeguards in AI Agricultural Advisory Services in Bangladesh
AI-supported agricultural advice raises questions about useful information, the responsibilities of providers and the accessibility of legal-rights information. This study examines stated preferences about these issues in 385 adult, AI-aware survey records from Bangladesh. The mixed sample consists of law students, graduates, legal professionals and other respondents; the workbook does not identify farmers, agricultural AI users or actual knowledge of farmers’ legal rights. The analysis therefore examines questionnaire responses about rights information and safeguards rather than claiming to measure farmers’ rights awareness or the effects of an AI service. Individual items are reported using descriptive distributions, paired comparisons and exploratory rank correlations with Holm adjustment. Of 385 respondents, 338 (87.8%) endorsed including legal-rights information in responsible digital agricultural services, 330 (85.7%) endorsed identifying responsibility for important advice, 325 (84.4%) endorsed communicating uncertainty and limitations, and 247 (64.2%) endorsed human assistance when AI advice is inadequate. The rights-information item exceeded the usefulness item by 24.4 percentage points; 215 respondents endorsed both. The statements differ in content, and their paired comparison does not identify the effects of a service or rank the value of its features. Exploratory correlations with the rights-information item were small and did not provide clear evidence of association after adjustment for four tests. The article distinguishes information about rights from demonstrated legal knowledge and recommends evaluation with identified agricultural users, specific legal scenarios and observable referral pathways.
- 16 November 2026
- Published by Springer Nature in Research Square
Open AccessFarmers Rights Information and Safeguards in AI Agricultural Advisory Services in Bangladesh
AI-supported agricultural advice raises questions about useful information, the responsibilities of providers and the accessibility of legal-rights information. This study examines stated preferences about these issues in 385 adult, AI-aware survey records from Bangladesh. The mixed sample consists of law students, graduates, legal professionals and other respondents; the workbook does not identify farmers, agricultural AI users or actual knowledge of farmers’ legal rights. The analysis therefore examines questionnaire responses about rights information and safeguards rather than claiming to measure farmers’ rights awareness or the effects of an AI service. Individual items are reported using descriptive distributions, paired comparisons and exploratory rank correlations with Holm adjustment. Of 385 respondents, 338 (87.8%) endorsed including legal-rights information in responsible digital agricultural services, 330 (85.7%) endorsed identifying responsibility for important advice, 325 (84.4%) endorsed communicating uncertainty and limitations, and 247 (64.2%) endorsed human assistance when AI advice is inadequate. The rights-information item exceeded the usefulness item by 24.4 percentage points; 215 respondents endorsed both. The statements differ in content, and their paired comparison does not identify the effects of a service or rank the value of its features. Exploratory correlations with the rights-information item were small and did not provide clear evidence of association after adjustment for four tests. The article distinguishes information about rights from demonstrated legal knowledge and recommends evaluation with identified agricultural users, specific legal scenarios and observable referral pathways.
- Open Access
- 13 November 2026
- Published by Springer Nature in Research Square
Patient Redress for AI-Related Medical Errors in Bangladesh: A Socio-Legal Survey
Responsibility for artificial intelligence (AI) assisted medical decisions encompasses both preventing harm and providing a way to respond when harm occurs. These functions are often combined in empirical measures of accountability, making it difficult to identify which safeguards respondents prioritise. This cross-sectional analysis used 385 adult responses to a common questionnaire on AI, health law and patient rights in Bangladesh. Three preventive statements addressed defined responsibility, independent clinical judgement and organisational and developer safety controls. A separate statement addressed access to complaints and remedies; a fifth elicited anticipated trust under clear accountability. Endorsement was compared within respondents using exact McNemar tests with Holm adjustment. Exploratory regression distinguished prevention and redress as correlates of anticipated trust. Redress was endorsed by 311 respondents (80.8%), exceeding endorsement of each preventive statement by 16.6–20.5 percentage points (all adjusted p < 0.001). The adjusted linear association with anticipated trust was positive for the prevention summary (b = 0.205, 95% CI 0.049 to 0.361; adjusted p = 0.020), but attenuated in inference when trust was dichotomised. The redress coefficient was close to zero in the adjusted linear model (b = 0.001, 95% CI −0.112 to 0.114). These findings distinguish broad endorsement of a patient-facing remedy from covariance with a hypothetical trust judgement. They support evaluating complaint access as a governance outcome in its own right. The survey neither verifies actual use of medical AI nor estimates effects of introducing safeguards.
- 13 November 2026
- Published by Springer Nature in Research Square
Open AccessPatient Redress for AI-Related Medical Errors in Bangladesh: A Socio-Legal Survey
Responsibility for artificial intelligence (AI) assisted medical decisions encompasses both preventing harm and providing a way to respond when harm occurs. These functions are often combined in empirical measures of accountability, making it difficult to identify which safeguards respondents prioritise. This cross-sectional analysis used 385 adult responses to a common questionnaire on AI, health law and patient rights in Bangladesh. Three preventive statements addressed defined responsibility, independent clinical judgement and organisational and developer safety controls. A separate statement addressed access to complaints and remedies; a fifth elicited anticipated trust under clear accountability. Endorsement was compared within respondents using exact McNemar tests with Holm adjustment. Exploratory regression distinguished prevention and redress as correlates of anticipated trust. Redress was endorsed by 311 respondents (80.8%), exceeding endorsement of each preventive statement by 16.6–20.5 percentage points (all adjusted p < 0.001). The adjusted linear association with anticipated trust was positive for the prevention summary (b = 0.205, 95% CI 0.049 to 0.361; adjusted p = 0.020), but attenuated in inference when trust was dichotomised. The redress coefficient was close to zero in the adjusted linear model (b = 0.001, 95% CI −0.112 to 0.114). These findings distinguish broad endorsement of a patient-facing remedy from covariance with a hypothetical trust judgement. They support evaluating complaint access as a governance outcome in its own right. The survey neither verifies actual use of medical AI nor estimates effects of introducing safeguards.
- Open Access
- 13 November 2026
- Published by Springer Nature in Research Square
Intellectual Property Safeguards and the Role of Fairness in Acceptance of AI Generated Content
Public endorsement of safeguards for AI-generated creative content does not establish how strongly fairness figures in personal acceptance of that content. This cross-sectional study distinguishes these judgements using 385 adult questionnaire records from Bangladesh. We analysed four statements on creator protection, transparent balancing rules, complaint access and transparency about data and content practices, and a fifth statement on the relevance of fairness to acceptance. We estimated full response distributions, joint endorsement and paired contrasts, followed by exploratory regressions of the original fairness-salience rating. Safeguard endorsement ranged from 78.2% to 80.3%, compared with 58.4% for the salience statement. Within-person differences ranged from 19.7 to 21.8 percentage points (all Holm-adjusted p < 0.001). Of the 301 respondents endorsing access to complaints, 176 also endorsed the salience statement. A copyright-oriented summary was associated with a higher salience rating in the adjusted model (b = 0.358, 95% CI 0.181 to 0.536; Holm-adjusted p < 0.001). Reported production precautions and the governance expectation summary did not meet the adjusted threshold. The association with copyright orientation persisted when the outcome was dichotomised and when the analysis was restricted to law students. The findings distinguish support for institutional protections from the reported place of fairness in personal judgement. They do not measure actual acceptance, legal compliance or the effect of changing a safeguard. The sample is concentrated in legal education and practice and does not establish national prevalence.
- 13 November 2026
- Published by Springer Nature in Research Square
Open AccessIntellectual Property Safeguards and the Role of Fairness in Acceptance of AI Generated Content
Public endorsement of safeguards for AI-generated creative content does not establish how strongly fairness figures in personal acceptance of that content. This cross-sectional study distinguishes these judgements using 385 adult questionnaire records from Bangladesh. We analysed four statements on creator protection, transparent balancing rules, complaint access and transparency about data and content practices, and a fifth statement on the relevance of fairness to acceptance. We estimated full response distributions, joint endorsement and paired contrasts, followed by exploratory regressions of the original fairness-salience rating. Safeguard endorsement ranged from 78.2% to 80.3%, compared with 58.4% for the salience statement. Within-person differences ranged from 19.7 to 21.8 percentage points (all Holm-adjusted p < 0.001). Of the 301 respondents endorsing access to complaints, 176 also endorsed the salience statement. A copyright-oriented summary was associated with a higher salience rating in the adjusted model (b = 0.358, 95% CI 0.181 to 0.536; Holm-adjusted p < 0.001). Reported production precautions and the governance expectation summary did not meet the adjusted threshold. The association with copyright orientation persisted when the outcome was dichotomised and when the analysis was restricted to law students. The findings distinguish support for institutional protections from the reported place of fairness in personal judgement. They do not measure actual acceptance, legal compliance or the effect of changing a safeguard. The sample is concentrated in legal education and practice and does not establish national prevalence.
- Open Access
- 12 November 2026
- Published by Springer Nature in Research Square
Trust Conditions and Regulatory Protection in AI Based Agricultural Information in Bangladesh
AI-supported agricultural information may provide useful guidance while raising concerns about local suitability, accountability and oversight. This study examines stated trust conditions and perceptions of regulatory protection in a Bangladesh survey. The 385 adult, AI-aware records comprise law students, graduates, legal professionals and other respondents; 60 report a location classified as upazila or rural. The file does not identify farmers or establish that respondents use agricultural AI, so findings concern a mixed sample rather than representative rural users. The analysis retains separate questionnaire items and compares paired endorsement of statements on source/provider clarity, regulatory oversight and accountability. It also explores correlations between AI awareness items and two trust-related statements. Of 385 respondents, 310 (80.5%) agreed that clear sources and a responsible provider would increase trust, 255 (66.2%) agreed that regulatory oversight can increase confidence, and 317 (82.3%) agreed that strong accountability increases willingness to use agricultural AI. Source/provider clarity exceeded oversight endorsement by 14.3 percentage points; 210 respondents endorsed both statements. These are differently worded hypothetical propositions, so their contrast is not a ranking of interventions or evidence of causal effects. Associations between awareness and trust conditions were small and do not establish that training changes reliance. The study contributes an item-level distinction between provider transparency, institutional oversight, local adaptation and accountability, and proposes a future design that evaluates specific advisory systems with actual agricultural users.
- 12 November 2026
- Published by Springer Nature in Research Square
Open AccessTrust Conditions and Regulatory Protection in AI Based Agricultural Information in Bangladesh
AI-supported agricultural information may provide useful guidance while raising concerns about local suitability, accountability and oversight. This study examines stated trust conditions and perceptions of regulatory protection in a Bangladesh survey. The 385 adult, AI-aware records comprise law students, graduates, legal professionals and other respondents; 60 report a location classified as upazila or rural. The file does not identify farmers or establish that respondents use agricultural AI, so findings concern a mixed sample rather than representative rural users. The analysis retains separate questionnaire items and compares paired endorsement of statements on source/provider clarity, regulatory oversight and accountability. It also explores correlations between AI awareness items and two trust-related statements. Of 385 respondents, 310 (80.5%) agreed that clear sources and a responsible provider would increase trust, 255 (66.2%) agreed that regulatory oversight can increase confidence, and 317 (82.3%) agreed that strong accountability increases willingness to use agricultural AI. Source/provider clarity exceeded oversight endorsement by 14.3 percentage points; 210 respondents endorsed both statements. These are differently worded hypothetical propositions, so their contrast is not a ranking of interventions or evidence of causal effects. Associations between awareness and trust conditions were small and do not establish that training changes reliance. The study contributes an item-level distinction between provider transparency, institutional oversight, local adaptation and accountability, and proposes a future design that evaluates specific advisory systems with actual agricultural users.
- Open Access
- 11 November 2026
- Published by Springer Nature in Research Square
Trust Conditions and Conditional Willingness to Use AI Legal Information in Bangladesh
A person can express willingness to use an artificial intelligence (AI) legal service under transparent conditions while remaining cautious about acting on its information. This cross-sectional study distinguishes conditional willingness, trust expectations and verification caution using 385 adult questionnaire records from Bangladesh. We report full response distributions, paired comparisons and exploratory adjusted models of willingness attributed to transparent explanations. The willingness statement was endorsed by 337 respondents (87.5%); trust conditional on clear sources was endorsed by 260 (67.5%) and trust conditional on qualified professional oversight by 266 (69.1%). Within respondents, willingness endorsement exceeded these two trust conditions by 20.0 and 18.4 percentage points, respectively (both Holm-adjusted p < 0.001). Of the 337 respondents who endorsed conditional willingness, 233 also endorsed the professional oversight condition. In the adjusted original-rating model, perceived access utility (b = 0.275, 95% CI 0.111 to 0.439) and verification caution (b = 0.154, 95% CI 0.060 to 0.248) were positively associated with conditional willingness (both adjusted p = 0.003). Neither association met the adjusted threshold when willingness was dichotomised. The findings distinguish openness to a transparent service from confidence in an answer or readiness to act on it. They motivate separate evaluation of source checking, professional escalation and service use. The study does not identify the causal effect of transparency or estimate actual adoption or national prevalence.
- 11 November 2026
- Published by Springer Nature in Research Square
Open AccessTrust Conditions and Conditional Willingness to Use AI Legal Information in Bangladesh
A person can express willingness to use an artificial intelligence (AI) legal service under transparent conditions while remaining cautious about acting on its information. This cross-sectional study distinguishes conditional willingness, trust expectations and verification caution using 385 adult questionnaire records from Bangladesh. We report full response distributions, paired comparisons and exploratory adjusted models of willingness attributed to transparent explanations. The willingness statement was endorsed by 337 respondents (87.5%); trust conditional on clear sources was endorsed by 260 (67.5%) and trust conditional on qualified professional oversight by 266 (69.1%). Within respondents, willingness endorsement exceeded these two trust conditions by 20.0 and 18.4 percentage points, respectively (both Holm-adjusted p < 0.001). Of the 337 respondents who endorsed conditional willingness, 233 also endorsed the professional oversight condition. In the adjusted original-rating model, perceived access utility (b = 0.275, 95% CI 0.111 to 0.439) and verification caution (b = 0.154, 95% CI 0.060 to 0.248) were positively associated with conditional willingness (both adjusted p = 0.003). Neither association met the adjusted threshold when willingness was dichotomised. The findings distinguish openness to a transparent service from confidence in an answer or readiness to act on it. They motivate separate evaluation of source checking, professional escalation and service use. The study does not identify the causal effect of transparency or estimate actual adoption or national prevalence.
- Open Access
- 11 November 2026
- Published by Springer Nature in Research Square
Patient Rights Safeguards and Conditional Acceptance of AI Assisted Diagnosis in Bangladesh
Acceptance of artificial intelligence in diagnosis raises questions about what people consider useful and which patient safeguards they endorse. This study examines their relationship in 385 adult, AI-aware questionnaire records supplied as empirical data from Bangladesh. Five diagnosis-and-rights statements are analysed separately because their content and low block reliability do not support a validated common acceptance scale. Response distributions, paired endorsement contrasts, eight combinations of rights preferences, Spearman correlations and exploratory logistic models distinguish professional usefulness, information, questioning, human judgement and conditional acceptance. The statement that patient-rights safeguards increase acceptance received 81.6% endorsement, compared with 56.9% for usefulness when AI supports qualified professionals. Information about material AI involvement received 81.0%, opportunities to ask questions 57.7%, and continuing human clinical judgement 63.9%. All three specific safeguards were jointly endorsed by 32.7%. Information preference had a small positive rank association with conditional acceptance after Holm correction. Adjusted information and questioning coefficients were positive but uncertain after correction and sensitivity checks. The findings identify differentiated endorsement of safeguards rather than an observed effect of rights protection on service uptake. They suggest evaluating communication as an opportunity for participation, not merely notification, while keeping normative rights distinct from an adoption strategy. Mixed respondent status, absent verification of patient experience, undocumented recruitment and fieldwork dates, and unresolved Codebook provenance labels restrict generalisation. The study provides an item-specific basis for further research connecting patient communication, professional oversight and conditional acceptance without claiming clinical effectiveness or tested legal knowledge.
- 11 November 2026
- Published by Springer Nature in Research Square
Open AccessPatient Rights Safeguards and Conditional Acceptance of AI Assisted Diagnosis in Bangladesh
Acceptance of artificial intelligence in diagnosis raises questions about what people consider useful and which patient safeguards they endorse. This study examines their relationship in 385 adult, AI-aware questionnaire records supplied as empirical data from Bangladesh. Five diagnosis-and-rights statements are analysed separately because their content and low block reliability do not support a validated common acceptance scale. Response distributions, paired endorsement contrasts, eight combinations of rights preferences, Spearman correlations and exploratory logistic models distinguish professional usefulness, information, questioning, human judgement and conditional acceptance. The statement that patient-rights safeguards increase acceptance received 81.6% endorsement, compared with 56.9% for usefulness when AI supports qualified professionals. Information about material AI involvement received 81.0%, opportunities to ask questions 57.7%, and continuing human clinical judgement 63.9%. All three specific safeguards were jointly endorsed by 32.7%. Information preference had a small positive rank association with conditional acceptance after Holm correction. Adjusted information and questioning coefficients were positive but uncertain after correction and sensitivity checks. The findings identify differentiated endorsement of safeguards rather than an observed effect of rights protection on service uptake. They suggest evaluating communication as an opportunity for participation, not merely notification, while keeping normative rights distinct from an adoption strategy. Mixed respondent status, absent verification of patient experience, undocumented recruitment and fieldwork dates, and unresolved Codebook provenance labels restrict generalisation. The study provides an item-specific basis for further research connecting patient communication, professional oversight and conditional acceptance without claiming clinical effectiveness or tested legal knowledge.