On inscription and bias: data, actor network theory, and the social problems of text-to-image AI models

Abstract

Text-to-image generation platforms are a type of generative artificial intelligence that can produce novel and realistic images from a text prompt. However, these systems also raise social and ethical issues related to the data they rely on. Therefore, this review essay explores how data influence these issues and how to address them using the concept of inscription by Bruno Latour. Inscription is the process of encoding the values and interests of the actors involved in the creation and use of a technology into the technology itself. Using inscription as a theoretical and analytical tool, this work analyzes the data sources, data processing, data representation, and data interpretation of these systems, and reveals how they shape the images they generate and the potential biases and harms they may cause. Thus, this essay offers a new perspective on the ethical discussion of the generative AI models, especially text-to-image models, by bridging the gap between the technical and sociological perspectives on these issues, which has been largely overlooked in the existing literature, and it also provides some novel and practical recommendations for the developers, users, and regulators of these technologies, based on the findings and implications of the analysis.

On inscription and bias: data, actor network theory, and the social problems of text-to-image AI models

Abstract

Text-to-image generation platforms are a type of generative artificial intelligence that can produce novel and realistic images from a text prompt. However, these systems also raise social and ethical issues related to the data they rely on. Therefore, this review essay explores how data influence these issues and how to address them using the concept of inscription by Bruno Latour. Inscription is the process of encoding the values and interests of the actors involved in the creation and use of a technology into the technology itself. Using inscription as a theoretical and analytical tool, this work analyzes the data sources, data processing, data representation, and data interpretation of these systems, and reveals how they shape the images they generate and the potential biases and harms they may cause. Thus, this essay offers a new perspective on the ethical discussion of the generative AI models, especially text-to-image models, by bridging the gap between the technical and sociological perspectives on these issues, which has been largely overlooked in the existing literature, and it also provides some novel and practical recommendations for the developers, users, and regulators of these technologies, based on the findings and implications of the analysis.

The persuasive effects of social cues and source effects on misinformation susceptibility

Abstract

Although misinformation exposure takes place within a social context, significant conclusions have been drawn about misinformation susceptibility through studies that largely examine judgements in a social vacuum. Bridging the gap between social influence research and the cognitive science of misinformation, we examine the mechanisms through which social context impacts misinformation susceptibility across 5 experiments (N = 20,477). We find that social cues only impact individual judgements when they influence perceptions of wider social consensus, and that source similarity only biases news consumers when the source is high in credibility. Specifically, high and low engagement cues (‘likes’) reduced misinformation susceptibility relative to a control, and endorsement cues increased susceptibility, but discrediting cues had no impact. Furthermore, political ingroup sources increased susceptibility if the source was high in credibility, but political outgroup sources had no effect relative to a control. This work highlights the importance of studying cognitive processes within a social context, as judgements of (mis)information change when embedded in the social world. These findings further underscore the need for multifaceted interventions that take account of the social context in which false information is processed to effectively mitigate the impact of misinformation on the public.

The persuasive effects of social cues and source effects on misinformation susceptibility

Abstract

Although misinformation exposure takes place within a social context, significant conclusions have been drawn about misinformation susceptibility through studies that largely examine judgements in a social vacuum. Bridging the gap between social influence research and the cognitive science of misinformation, we examine the mechanisms through which social context impacts misinformation susceptibility across 5 experiments (N = 20,477). We find that social cues only impact individual judgements when they influence perceptions of wider social consensus, and that source similarity only biases news consumers when the source is high in credibility. Specifically, high and low engagement cues (‘likes’) reduced misinformation susceptibility relative to a control, and endorsement cues increased susceptibility, but discrediting cues had no impact. Furthermore, political ingroup sources increased susceptibility if the source was high in credibility, but political outgroup sources had no effect relative to a control. This work highlights the importance of studying cognitive processes within a social context, as judgements of (mis)information change when embedded in the social world. These findings further underscore the need for multifaceted interventions that take account of the social context in which false information is processed to effectively mitigate the impact of misinformation on the public.

Beliefs Matter: Local Climate Concerns and Industrial Greenhouse Gas Emissions in the United States

Abstract

Industrial emissions of greenhouse gases are significant contributors to climate change, which poses a grave threat to social and economic systems. Our understanding of what might drive firms to reduce their emissions of these gases, however, is incomplete, and it is not clear that the knowledge gained from other environmental issues will readily apply to these emissions. We argue and find that indicators of environmental injustice previously shown to relate to toxic pollutants, for example, are poor predictors of greenhouse gas emissions. Instead, we show that the degree of belief in and concern about climate change in a local community is a significant predictor of the facility’s rate of emission improvements. Furthermore, we find that beliefs at both the facility and headquarter communities influence emission reduction, and that those effects are substitutes for each other.

Beliefs Matter: Local Climate Concerns and Industrial Greenhouse Gas Emissions in the United States

Abstract

Industrial emissions of greenhouse gases are significant contributors to climate change, which poses a grave threat to social and economic systems. Our understanding of what might drive firms to reduce their emissions of these gases, however, is incomplete, and it is not clear that the knowledge gained from other environmental issues will readily apply to these emissions. We argue and find that indicators of environmental injustice previously shown to relate to toxic pollutants, for example, are poor predictors of greenhouse gas emissions. Instead, we show that the degree of belief in and concern about climate change in a local community is a significant predictor of the facility’s rate of emission improvements. Furthermore, we find that beliefs at both the facility and headquarter communities influence emission reduction, and that those effects are substitutes for each other.

Naive skepticism scale: development and validation tests applied to the chilean population

Abstract

Background

Skepticism has traditionally been associated with critical thinking. However, philosophy has proposed a particular type of skepticism, termed naive skepticism, which may increase susceptibility to misinformation, especially when contrasting information from official sources. While some scales propose to measure skepticism, they are scarce and only measure specific topics; thus, new instruments are needed to assess this construct.

Objective

This study aimed to develop a scale to measure naive skepticism in the adult population.

Method

The study involved 446 individuals from the adult population. Subjects were randomly selected for either the pilot study (phase 2; n = 126) or the validity-testing study (phase 3; n = 320). Parallel analyses and exploratory structural equation modelling were conducted to assess the internal structure of the test. Scale reliability was estimated using Cronbach's alpha and McDonald's omega coefficients Finally, a multigroup confirmatory factor analysis was performed to assess invariance, and a Set- Exploratory Structural Equation Modeling was applied to estimate evidence of validity based on associations with other variables.

Results

The naive skepticism scale provided adequate levels of reliability (ω > 0.8), evidence of validity based on the internal structure of the test (CFI = 0.966; TLI = 0.951; RMSEA = 0.079), gender invariance, and a moderate inverse effect on attitudes towards COVID-19 vaccines.

Conclusions

The newly developed naive skepticism scale showed acceptable psychometric properties in an adult population, thus enabling the assessment of naive skepticism in similar demographics. This paper discusses the implications for the theoretical construct and possible limitations of the scale.

Naive skepticism scale: development and validation tests applied to the chilean population

Abstract

Background

Skepticism has traditionally been associated with critical thinking. However, philosophy has proposed a particular type of skepticism, termed naive skepticism, which may increase susceptibility to misinformation, especially when contrasting information from official sources. While some scales propose to measure skepticism, they are scarce and only measure specific topics; thus, new instruments are needed to assess this construct.

Objective

This study aimed to develop a scale to measure naive skepticism in the adult population.

Method

The study involved 446 individuals from the adult population. Subjects were randomly selected for either the pilot study (phase 2; n = 126) or the validity-testing study (phase 3; n = 320). Parallel analyses and exploratory structural equation modelling were conducted to assess the internal structure of the test. Scale reliability was estimated using Cronbach's alpha and McDonald's omega coefficients Finally, a multigroup confirmatory factor analysis was performed to assess invariance, and a Set- Exploratory Structural Equation Modeling was applied to estimate evidence of validity based on associations with other variables.

Results

The naive skepticism scale provided adequate levels of reliability (ω > 0.8), evidence of validity based on the internal structure of the test (CFI = 0.966; TLI = 0.951; RMSEA = 0.079), gender invariance, and a moderate inverse effect on attitudes towards COVID-19 vaccines.

Conclusions

The newly developed naive skepticism scale showed acceptable psychometric properties in an adult population, thus enabling the assessment of naive skepticism in similar demographics. This paper discusses the implications for the theoretical construct and possible limitations of the scale.

Digital citizenship and its relevance for literacy education: perspectives of preservice teachers

Abstract

This research examines preservice teachers’ perspectives on digital citizenship and its relevance for literacy education. Digital citizenship has been explored in various ways in recent decades, primarily in the educational technology literature, and prominent themes of digital citizenship include the use of safely, responsibly, and productively using technology to engage in digital spaces. However, limited research has explored preservice teachers’ perspectives on digital citizenship, particularly as relates to literacy education. This qualitative case study investigates the views of 111 preservice teachers enrolled in a teacher preparation program in the Midwestern United States on digital citizenship and its relevance for literacy education. Data analysis of participants’ written reflections revealed four primary themes of digital citizenship that align with concepts, skills, and goals in literacy education: well-informed citizens, community engagement and activism, safety, and technological know-how. These themes also connect to digital literacies. Given the commonalities between digital citizenship and literacy education, explicit integration of digital citizenship curricula into literacy education can connect important and discrete digital literacy skills into more cohesive educational units that empower children to safely and productively utilize digital technologies to promote meaningful change in their communities.

Digital citizenship and its relevance for literacy education: perspectives of preservice teachers

Abstract

This research examines preservice teachers’ perspectives on digital citizenship and its relevance for literacy education. Digital citizenship has been explored in various ways in recent decades, primarily in the educational technology literature, and prominent themes of digital citizenship include the use of safely, responsibly, and productively using technology to engage in digital spaces. However, limited research has explored preservice teachers’ perspectives on digital citizenship, particularly as relates to literacy education. This qualitative case study investigates the views of 111 preservice teachers enrolled in a teacher preparation program in the Midwestern United States on digital citizenship and its relevance for literacy education. Data analysis of participants’ written reflections revealed four primary themes of digital citizenship that align with concepts, skills, and goals in literacy education: well-informed citizens, community engagement and activism, safety, and technological know-how. These themes also connect to digital literacies. Given the commonalities between digital citizenship and literacy education, explicit integration of digital citizenship curricula into literacy education can connect important and discrete digital literacy skills into more cohesive educational units that empower children to safely and productively utilize digital technologies to promote meaningful change in their communities.