The psychology of artificial intelligence trustworthiness

Enrique Leon, Artificial Intelligence and Cloud Enterprise Architect at American Sugar Refining

Artificial intelligence (AI) is increasingly used to generate content, such as text, images, music, and videos, that can influence human beliefs, attitudes, and behaviors. However, not all AI-generated content is accurate, reliable or ethical. Some AI systems may intentionally or unintentionally produce misleading, biased, or harmful content, which may have negative consequences for individuals and society. Therefore, it is important to understand how people assess the trustworthiness of AI-generated content and how it compares to human-generated content.

This article explores the psychological factors that influence people’s trust in AI-generated content and why they are more likely to accept the authenticity of AI-generated content than human-generated content. We review the existing literature on this topic and propose a conceptual framework to explain the main cognitive and affective processes involved. We also discuss the implications of our findings for the design and regulation of AI systems and the education and empowerment of users.

literature review

A growing body of research explores how people perceive and respond to AI-generated content, particularly in the field of text and image generation. Some of the main themes that emerge from this literature are:

• People generally tend to trust content generated by AI, especially if they are unaware of its source or have a positive attitude toward AI.

• People are influenced by the quality, coherence and consistency of AI-generated content and the cues and context that come with it.

• People are more likely to accept AI-generated content when it confirms their prior beliefs, preferences, or expectations, or when it appeals to their emotions or motivations.

• People are less likely to question or verify AI-generated content than human-generated content because they have a lower perception of responsibility, responsibility, or intentionality from the AI ​​source.

• People are more susceptible to AI-generated content when they have low levels of media literacy, critical thinking or digital skills, or when they are in situations of high uncertainty, complexity or information overload.

conceptual framework

Based on a literature review, we propose a conceptual framework that illustrates the main psychological factors that influence trust in AI-generated content and how they compare to human-generated content. This framework consists of four components: source, message, recipient, and context. Each component has several subcomponents that represent specific variables that influence people’s trust. The framework ends with interactions and feedback loops between components and subcomponents.

“Users should be empowered and involved in the co-creation and governance of artificial intelligence systems and have the opportunity to express their opinions and concerns about the system and its outputs”

Conceptual Framework in the Psychology of Artificial Intelligence Credibility Perceived Objectivity – Artificial intelligence is simply perceived to be objective.

Consistency and reliability—trust based on consistent and high-quality content

Authoritative Attribution – Artificial Intelligence uses advanced technology, and most people don’t realize that Artificial Intelligence goes back decades

Lack of emotional bias – Artificial intelligence lacks emotions, thereby reducing the concerns associated with these emotions.Transparency—Trust is achieved through transparent explanations of user perceptions

Accuracy and precision – users trust that artificial intelligence is accurate and precise

Social Proof – Widespread adoption of AI and positive user experience

Mitigating Confirmation Bias – Content can mitigate confirmation bias by presenting information objectively

discuss

The conceptual framework I propose can help us understand the psychological mechanisms behind people’s trust in AI-generated content and why they are more accepting of its authenticity than human-generated content. The framework can also inform the design and regulation of AI systems and the education and empowerment of users. Some possible impacts are:

• AI systems should be transparent and accountable about their sources, methods, and goals, and provide clear and accurate information about the quality, reliability, and limitations of their outputs.

• AI systems should be ethical and responsible in producing content that respects human values, rights and dignity and avoid misleading, biased or harmful content.

• AI systems should be able to adapt and respond to user feedback and preferences, allowing users to control and customize their interactions with the system.

• Users should understand and understand the existence and potential impact of AI-generated content and develop the skills and abilities to critically evaluate and validate the content they encounter.

• Users should be empowered and involved in the co-creation and governance of AI systems and have the opportunity to express their opinions and concerns about the system and its outputs.

In this article, we explore the psychology of AI trustworthiness and why people trust AI-generated content more than human-generated content. We review the existing literature on this topic and propose a conceptual framework to explain the main cognitive and affective processes involved. We also discuss the implications of our findings for the design and regulation of AI systems and for user education and empowerment. I hope that this article will contribute to the advancement of research and practice in this important emerging area.