Taking AI risks seriously: a new assessment model for the AI Act

Abstract

The EU Artificial Intelligence Act (AIA) defines four risk categories: unacceptable, high, limited, and minimal. However, as these categories statically depend on broad fields of application of AI, the risk magnitude may be wrongly estimated, and the AIA may not be enforced effectively. This problem is particularly challenging when it comes to regulating general-purpose AI (GPAI), which has versatile and often unpredictable applications. Recent amendments to the compromise text, though introducing context-specific assessments, remain insufficient. To address this, we propose applying the risk categories to specific AI scenarios, rather than solely to fields of application, using a risk assessment model that integrates the AIA with the risk approach arising from the Intergovernmental Panel on Climate Change (IPCC) and related literature. This integrated model enables the estimation of AI risk magnitude  by considering the interaction between (a) risk determinants, (b) individual drivers of determinants, and (c) multiple risk types. We illustrate this model using large language models (LLMs) as an example.

Child and Adolescent Engagement with Climate Change on Social Media and Impacts on Mental Health: a Narrative Review

Abstract

Purpose of the Review

We describe the existing literature which explores the relationship between engagement with climate change on social media and child and adolescent mental health and well-being.

Recent Findings

Children and adolescents use social media to gather information about climate change, build community with like-minded peers, and get involved in collective climate action. Climate anxiety can motivate young people to seek out climate-related information on social media.

Summary

Social media has benefits to child and adolescent mental health in the context of the climate crisis as a tool to promote awareness, social support, and climate-related civic engagement. However, social media can spread misinformation and increase child and adolescent exposure to negative climate change messaging, thereby increasing distress. Clinicians and young people alike recognize the advantages of social media for promoting child and adolescent resilience in response to climate change and offer suggestions for how to reduce potential harm.

Misinformation and higher-order evidence

Abstract

This paper uses computational methods to simultaneously investigate the epistemological effects of misinformation on communities of rational agents, while also contributing to the philosophical debate on ‘higher-order’ evidence (i.e. evidence that bears on the quality and/or import of one’s evidence). Modelling communities as networks of individuals, each with a degree of belief in a given target proposition, it simulates the introduction of unreliable mis- and disinformants, and records the epistemological consequences for these communities. First, using small, artificial networks, it compares the effects, when agents who are aware of the prevalence of mis- or disinformation in their communities, either deny the import of this higher-order evidence, or attempt to accommodate it by distrusting the information in their environment. Second, deploying simulations on a large(r) real-world network, it observes the impact of increasing levels of misinformation on trusting agents, as well as of more minimal, but structurally targeted, unreliability. Comparing the two information processing strategies in an artificial context, it finds that there is a (familiar) trade-off between accuracy (in arriving at a correct consensus) and efficiency (in doing so in a timely manner). And in a more realistic setting, community confidence in the truth is seen to be depressed in the presence of even minimal levels of misinformation.

Barriers to correct pronoun usage in healthcare settings

Abstract

Background

Using correct pronouns is an impactful way to establish affirming environments for transgender and nonbinary (TGNB) patients. However, physicians often report struggling with this.

Objective

This study set out to conduct an initial root cause analysis of factors contributing to medical students and physicians failing to use TGNB patients’ correct pronouns.

Methods

A 10-item Qualtrics survey was sent to medical students, residents, and physicians practicing in Central Ohio. Participants were asked to describe perceived challenges or barriers colleagues have regarding correctly using TGNB patients’ correct pronouns. A directed content analysis of participant responses was performed utilizing a fishbone diagram root cause analysis tool as a basis for conceptualizing and categorizing barriers. All coding was completed by independent reviewers utilizing a consensus reconciliation methodology.

Results

Of 928 survey respondents, 763 met the study inclusion criteria, of which 453 provided analyzable responses. Of these 453, attendings with five or more years of practice (32.5%) and medical students (27.4%) made up the two largest demographic categories. 1.7% of respondents identified as transgender, nonbinary, and/or genderqueer, and 64% identified as heterosexual/straight. Five core barrier categories were identified: documentation, patient care, environment, knowledge, and individuals. Sub-categories were also identified, including lack of documentation, discomfort, medical culture, lack of standardization, prejudice, and assumptions.

Conclusion

The study identifies important barriers to medical professionals correctly using TGNB patients’ pronouns. The root cause analysis conducted as part of this study demonstrates the necessity of multi-pronged, system-level interventions to support ensuring TGNB patients are addressed using the correct pronouns.

From crisis to prevention: mining big data for public health insights during the flint water crisis

Abstract

This study investigates the utility of Google Trends and Google Search data in retrospectively analyzing the Flint Water Crisis, a significant public health event. By examining keywords relevant to the crisis, such as “lead,” “bottled water,” “water filters,” “pneumonia,” and “Legionnaires’ Disease,” we aimed to uncover patterns in public awareness and response during a public crisis and explore how such awareness and responses could potentially aid in reducing the risk of similar crises in the future. Our analysis reveals a clear correlation between search frequencies and the crisis timeline, with spikes in search terms corresponding to key events. This suggests that such data can serve as a valuable tool for understanding public sentiment and behavior in the face of environmental disasters. The research underscores the potential of “Big Data,” led by search engines and social media platforms, in shaping public policy and informed decision-making. However, it also addresses the limitations and challenges in using these data sources, including issues of data consistency, replicability, and the influence of sociocultural contexts on search behaviors. The study advocates for the combined use of Google Trends and Google Search data, complemented by other datasets, for a more comprehensive understanding of public engagement in environmental crises. This work contributes to the growing field of infodemiology, emphasizing the importance of big data analysis in environmental science and public health research.

Global citizenship identity mediates the relationship of knowledge, cognitive, and socio-emotional skills with engagement towards global issues

Abstract

The UN Sustainable Development Goals (SDGs) provide a unified framework to address interconnected global issues, emphasizing the need for collective action across all sectors of society to achieve a sustainable future for all. In this paper, we empirically investigate how knowledge (awareness of global issues), cognitive skills (critical inquiry), and socio-emotional skills (cognitive empathy) relate to engagement towards global issues, and whether global citizenship identification mediates these relationships. Mediation analysis of data from 249 participants revealed that both awareness of global issues and cognitive empathy directly predict higher engagement levels. In contrast, no direct effect of critical inquiry was observed. Global citizenship identification significantly mediated the relationships between all three predictors and engagement: accounting for 70.7% of the effect of critical inquiry, 39.9% of the effect of awareness, and 33.6% of the effect of cognitive empathy. Our findings highlight that global citizenship identification plays a crucial role in translating knowledge and skills into active engagement. The results highlight the potential effectiveness of identity-based interventions in fostering more engaged communities and advancing efforts toward achieving the SDGs.

Towards gender-transformative metrics in seed system performance measurement: insights for policy and practice in Sub-Sahara Africa

Abstract

Context

Food insecurity in Sub-Sahara Africa hinges on addressing salient gender inequities within the seed system. While efficient seed system promises reduced systemic inefficiencies to fast-track seed delivery to the smallholder farmers, a dearth of standardized industry metrices to understand the intersectionality of seed system and gender issues exist. Specifically, metrices on guaranteed seed access, reach, benefit, women’s empowerment and ultimate transformation of women, youth and vulnerable people’s livelihoods are less understood. The existing metrices are aggregated at very high levels and limit the ability of policymakers and industry stakeholders to effectively address gender-based inequities for an optimized seed system.

Objective

Our objective is to challenge the status quo industry metrics used by seed industry players and apply a gender framework that strikes a balance between the needs of women, youth and vulnerable peoples in the system, vis-a-vis the need of public, private, and civil society actors. Therefore, the study seeks to evaluate how seed system metrics can be effectively tailored to address gender gaps for enhanced agricultural productivity and food security in Sub-Sahara African context. It also refines the proposals of Kennedy and Speilman and introduce gender-specific metrices that may hold promise to address women and youth’s challenges within the seed system.

Methods

A systemic review of current industry metrices was conducted and the newly developed reach, benefit, empower and transform (RBET) framework was applied to synthesize the responsiveness of current seed industry indicators on gender issues. Online databases and repositories with key search words that returned 204 results including some gray literature.

Results and conclusion

Using common bean seed system as an illustration, the study found critical gaps in measuring seed industry performance, innovation, structure, seed registration and quality control, intellectual property rights using the reach, benefit, empower and transform approach. Thus, a set of gender responsive indicators was suggested to address gender and inclusive matrices that the seed industry often neglects. Using the reach, benefit, empower and transform approach we have included gender responsive indicators meant to close existing gender gaps. Some of these indicators addressed include women participation, trait preferences, seed packaging sizing, seed system leadership, decision-making capacities, labor intensity/drudgery and use of digital platforms such as point-of-sale tracking systems to reach last mile farmers among others.

Significance

This study uses the newly–developed Reach, Benefit, Empower, and Transform (RBET) Framework together with the already existing Spielman–Kennedy framework. It is timely to inform policymaking process on seed system design, to enhance seed industry performance monitoring, and provide practitioners with the knowledge and missing links in efforts to align the seed system's performance with gender outcomes in a measurable manner.

An assessment of Ghana’s pilot of the RTS,S malaria vaccine implementation programme; 2019–2021: a retrospective study

Abstract

Background

In May 2019, Ghana piloted the introduction of RTS,S malaria vaccine into routine immunization in 42 districts of seven of the 16 regions. The RTS,S malaria vaccine implementation programme (MVIP) post-introduction evaluation (PIE) conducted in Ghana, assessed the immunization system as well as healthcare worker and caregiver experiences during the phase-one rollout but was less expressive on quantitative grading of the respective thematic areas of the vaccine introduction plan. Given the utility of summary statistics in programme evaluation and communication, this follow-up study aimed to provide an overall rating of the country's performance regarding the MVIP .

Methods

A retrospective study was conducted from 10th January to 5th February 2024. It involved review of records to assess key thematic areas of the national MVIP plan, using a study tool adapted from the WHO New Vaccine Introduction (NVI) checklist. A composite score ranging from zero to 100 per cent was generated to assess the country's overall performance regarding introduction of the malaria vaccine, rated on a Likert scale as comprehensive, good, fair, and poor.

Results

The overall performance in the MVIP was rated 78.9% (30/38) corresponding to a grading of “good” on the Likert scale. Performance indicators under thematic areas including policy, national coordination mechanisms, waste management, health worker training, and pharmacovigilance were completely achieved. However,  some weaknesses were exhibited in areas such as financial consideration, cold chain, logistics, and vaccine management, and monitoring and evaluation.

Conclusion

Ghana’s MVIP demonstrated remarkable strengths worth leveraging  to improve the national immunization programme. The weaknesses observed in some of the thematic areas present opportunities to engage key immunization partners and stakeholders towards aligning efforts to ensure a more robust expansion phase. The lessons from the MVIP may be relevant to areas introducing malaria vaccine irrespective of the product type—RTS,S or R21.

A qualitative study of knowledge, beliefs and misinformation regarding COVID-19 in selected districts in Zimbabwe

Abstract

Background

Lack of appropriate knowledge, incorrect beliefs and misinformation misleads people about the risks they face and how best to protect themselves. A study was conducted to explore the knowledge, beliefs and misinformation regarding COVID-19 in Zimbabwe.

Methods

A qualitative study was conducted in September-October 2022 with a purposive sample of religious leaders, women leaders, youth leaders, health workers, village health workers, teachers, traditional healers, transporters, and the general population selected from ten sites across the country. In total there were 128 participants (30 key informants and 98 focus group discussion participants). At each site, 3 key informant interviews and one homogenous focus group discussion were conducted using semi-structured interviews and focus group discussion guides, respectively. The data were recorded on audiotapes, transcribed verbatim, and translated into English. Manual thematic analysis of the data was performed.

Findings

Three themes were identified in this study: (1) beliefs about COVID-19, (2) knowledge about COVID-19 (knowledge of origin, definition, transmission, signs and symptoms and recommended preventive measures), and (3) misinformation about COVID-19 (regarding its nature, existence and recommended preventive measures). There was awareness of the origin, transmission, signs and symptoms of COVID-19 among the participants. Participants reported that Zimbabwean communities were conversant with public health measures such as maintaining social distancing, wearing masks, and maintaining hand hygiene. However, misinformation was also observed to have circulated among the communities.

Conclusion

Participants demonstrated good knowledge of COVID-19. However, the misinformation circulating in the country calls for the government to establish structures to monitor the legitimacy of information coming through different sources and invest in providing information through trusted sources. In any disease outbreak, the government should engage its citizenry to understand their knowledge, beliefs and any misinformation that might influence adherence to disease preventive measures.

Taking the public seriously: the role of respect in interactions between scientific experts and lay publics

Abstract

The way we engage with each other in science matters. While some ways of engaging may facilitate interactions, others may hinder them. Trust has been identified as one of the central factors facilitating collaborations between scientific communities and lay communities, and respect has been pointed to as having a central role to play in building and maintaining this trust. But what should respecting others in the interactions between scientific and lay communities involve? What does cultivating respect involve in this context? This paper aims at addressing these questions. In particular, it focuses on the role and nature of respect in the interactions between healthcare providers and patients in order to develop an analysis of the different types of respect and their relative importance in collaborations between scientists and lay publics. Overall, we will argue that understanding the complexity of respect dynamics may help to act appropriately in scientific experts-lay people interactions.