Aug 2026· AI & SOCIETY· 0 citations· 16 references
TL;DR
An understanding gap is illuminated: user attributions are partly guided by epistemic orientation and experiential ascriptions that make sophisticated simulation appear as understanding, especially in affective interaction, raising urgent questions about epistemic trust, relational vulnerability, and the ethics of AI companionship.
Abstract
The rapid proliferation of conversational AI systems has led many users to attribute understanding, emotional sensitivity, and experiential qualities to AI chatbots. While these systems produce highly fluent and contextually appropriate responses, whether such attributions reflect genuine understanding or conflate functional performance with cognitive capacity remains unclear. This paper examines this attribution by integrating a rigorous philosophical framework with empirical findings from chatbot users. We ground our analysis in an expanded ‘knowledge of causes’ view, according to which genuine understanding requires knowledge of dependence relations and the cognitive capacity to manipulate them through analogical, counterfactual, and abductive reasoning. On this account, understanding presupposes an underlying cognitive architecture, not merely correct outputs or behavioral adequacy. We surveyed 122 adult chatbot users about their perceptions of their chatbots’ understanding and their own epistemic beliefs. Participants rated AI functionality significantly higher than AI understanding, and cognitive capacities higher than affective ones, with particularly low ratings for affective understanding. Hierarchical regressions revealed domain-sensitive drivers: cognitive understanding attribution tracked perceived cognitive functionality and a ‘functionality implies understanding’ stance, whereas affective understanding was uniquely predicted by epistemic stance and attitudes toward AI subjective experience, beyond perceived affective functionality. These findings illuminate an understanding gap: user attributions are partly guided by epistemic orientation and experiential ascriptions that make sophisticated simulation appear as understanding, especially in affective interaction, raising urgent questions about epistemic trust, relational vulnerability, and the ethics of AI companionship.
The recent advancements in conversational artificial intelligence have developed systems that are able to mimic human conversation, which can create the perception of hearing or understanding from a machine. This situation poses significant psychological issues about the nature of how human understanding is generated with AI, even though it doesn't have a real brain. What are the psychological mechanisms in relation to a perceived understanding when engaging in AI conversation and what are the key interactional factors that help create this experience? A conceptual mixed-methods approach was used that involved the results of human–computer interaction research, social psychology, and computational linguistics. It involved secondary empirical literature analysis and modelling of perceived understanding as a function of language alignment, contextual coherence, personalization, response timing and anthropomorphic cues. The findings suggest that high linguistic alignment, contextual coherence and adaptive personalization are important factors for perceived understanding. Faster response times and anthropomorphic design attributes contribute further toward social presence and emotional interaction. Together, these factors create a sense of empathy and understanding, leading users to attribute awareness to an algorithmic system where none exists. Based on the above findings, the study concludes that perceived understanding in AI conversations is a psychologically constructed experience that is driven by interactional design, not by machine understanding. This not only boosts user engagement and trust but also introduces the risk of emotional over-attribution and reliance on AI systems.
Muhammad Muaavia Khalid, Binish Nawaz, Faiza Latif et al.· Pakistan Journal of Positive...· 0 citations
Artificial intelligence (AI) chatbots (e.g., ChatGPT) can communicate in strikingly humanlike ways. This has prompted many chatbot users to attribute psychological properties, including consciousness, to these systems. However, there is little scientific evidence that current AI chatbots are conscious. How, then, should we understand people’s consciousness attributions to chatbots? Are they merely metaphorical claims, or literal expressions of genuine beliefs? If these attributions lack evidential support, are users epistemically blameworthy for making them, or might they be epistemically innocent, yielding significant benefits otherwise unattainable? This paper offers a conceptual analysis of consciousness attributions to AI chatbots and develops a multidimensional taxonomy of the attitudes they may express, ranging from non-doxastic stances (e.g., pretence) to different forms of belief, including delusions. This taxonomy helps avoid conflations by showing that linguistically identical attributions can reflect importantly different attitudes and degrees of epistemic commitment to the proposition that chatbots are conscious. The taxonomy also provides a framework for empirical studies to operationalize and measure different forms of epistemic commitment to AI consciousness. Using this taxonomy, I argue that although some consciousness attributions to chatbots are epistemically benign, and even some irrational ones may be epistemically innocent, many others render the attributor epistemically blameworthy.
This article clarifies the concept definitions and evaluation criteria of understanding in cognitive psychology by combining classic theories and experimental evidence, and uses these criteria as the analytical framework for the performance of “similar understanding” in contemporary artificial intelligence systems.
Ruo Qin· Journal of Language, Culture...· 0 citations
It is argued that psychological competence should become a core consideration for model providers, deploying organizations, researchers, and regulators concerned with the real-world effects of human-facing AI systems.
M. Economides, Paul M. Sacher, Samuel Salzer et al.· 0 citations
Conversational information seeking (CIS) systems now generate explanations, but we still evaluate them using retrieval-focused metrics such as faithfulness, completeness, and source attribution. These checks are necessary, but they do not tell us whether a response helps users form a coherent mental model. Cognitive science treats understanding as an active construction guided by causal structure, coherence, and the organization of information. Evaluation should therefore test whether explanations support integration, inference, and retention. This paper aims to define the target form of communication and a way to assess it. We introduce Explanatory Narratives for CIS, which combine the organizing benefits of storytelling to enable explanation. We then propose ARIC, a cognitively grounded framework for evaluating explanatory narratives across four comprehension stages: Attention, Representation, Integration, and Consolidation. To demonstrate its value, we apply ARIC to human-authored explanatory narratives and show how stage-based analysis yields actionable diagnostic insights. This shifts CIS evaluation toward the question the IR community increasingly faces: whether system-generated explanations actually help users understand.
Vahid Sadiri Javadi, Sadia Naseer, Ali Ather et al.· International Conference on...· 0 citations
Artificial intelligence now shapes much of how information reaches people, who read it, and what they make of it. The environments it creates are probabilistic, often fluent without being grounded, and curated by systems whose workings stay hidden. Most research on this shift has gone to AI literacy, explainability, and the mechanics of human–AI interaction. Far less has gone to a prior question: how does human cognition itself change to cope? The paper addresses that question with the Human Cognitive Adaptation Framework (HCAF). One hundred thirty-one adults who use AI-generated and algorithmically curated content daily completed a 25-item, six-point inventory covering five facets—orientation, dimensional literacy, ambiguity tolerance, coherence recognition and critique, and relational navigation. The research examined reliability, the correlations among facets, and dimensionality. The full scale was highly reliable (α = .902) and the data factored cleanly (KMO = .841). One factor dominated: its eigenvalue of 7.93 carried 31.7% of the variance, and parallel analysis retained a single factor. The facets correlated strongly with one another (r = .46 to .69), which reads as one adaptive response rather than five separate skills. Within that single capacity, people recognized synthetic coherence more readily than they tolerated the uncertainty recognition exposes, so adaptation is not uniform across its parts. The research used these results to develop the HCAF, which treats adaptation as the work of staying oriented, judging coherence, and holding up under instability, and draws out what follows for human-centered AI, AI literacy, education, and the design of AI-supported decision environments.
C. Andoniou· 2026 International Conferenc...· 0 citations