Jul 2026· 2026 International Conference on Intelligent and Sustainable AI Systems (ICOSAAS)· pp. 544-550· 0 citations· 17 references
Abstract
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.
People increasingly reason with large language models (LLMs), yet complementary capabilities do not guarantee outperforming both components. In a between-subjects study, participants (N=535) solved a 40-item battery of matrix reasoning, mental rotation, syllogisms, and letter-string analogies, unaided or with GPT-5.6-Luna, Claude Opus 4.8, Gemini 3.6 Flash, or Kimi K3. Each assisted trial required consultation with the model. Each model answered every item alone 100 times under matched elicitation. The assisted-unaided accuracy difference increased with item-level LLM competence. Deference varied across tasks and increased with competence within tasks. Post-advice confidence distinguished correct from incorrect answers less strongly than unaided confidence. In a reference comparison, about half the increase in LLM accuracy carried through to assisted accuracy. How much of that accuracy gain reached participants differed across the models. These findings motivate evaluating LLMs in interaction with humans and designing support for selective deference that preserves independent reasoning.
Robin Welsch, Michelle Rausch, Pascal Knierim et al.· 0 citations
How can humans make sense of the rapid takeoff of artificial intelligence (AI)? We studied the sensemaking dynamics of AI through an open-ended, mixed-methods study with computational text analysis of millions of AI-related newspaper articles and social media posts grounded in 57 semi-structured interviews with AI professionals in 2021 and 2023--before and after the recent surge of public interest. We identify a range of sociological frames (interpretive schemas that structure collective cognition) and show how AI professionals use frames to address significant cognitive challenges, such as assigning responsibility for societal impacts. We develop a framework of three primary debates across which frames are adopted and contested: (i) the $\textit{method}$ of AI development, between frames of top-down expert systems and bottom-up emergent capabilities, (ii) the $\textit{mind}$ of an AI system, ranging from a passive tool to a humanlike"digital mind,"and (iii) the $\textit{morality}$ of how AI is used, particularly the decision of whether to slow down or speed up AI development. As humanity enters the era of transformative AI, technologists and policymakers must account for the framing dynamics that will circumscribe our beliefs, values, and actions.
Jacy Reese Anthis, Erik Brynjolfsson, James A. Evans· Proceedings of the ACM on Hu...· 0 citations
If AI is to support human cognitive growth, design must move beyond answer provision and efficiency maximization toward the organization of productive human-AI relations: relations that challenge users’ initial assumptions while providing support appropriate to the task and the user's level of expertise.
Xiaokun Wu, Min Chen, Giancarlo Fortino· Big Data and Cognitive Compu...· 0 citations
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
The Cognitive Reflection Test (CRT) is widely used in reasoning and decision making research, but it lacks a formal cognitive foundation. As a result, debates persist over why CRT scores correlate with mathematical ability, why some individuals solve CRT problems easily while others struggle, and which mental processes drive observed patterns in data. We use an ecological perspective combined with computational cognitive modeling to address these questions, focusing in particular on the bat-and-ball problem. First, we specify the learning environment by assembling a large dataset of grade school verbal math problems. Second, we specify a formal learning mechanism that selects the arithmetic operation most likely to be applied to a novel problem based on the linguistic association of the problem with learned exemplars. Our model generates the intuitive errors elicited by the bat-and-ball problem (as well as other CRT items) and explains why these errors can be seen as byproducts of adaptive cognition. It also makes new predictions about the effect of environmental structure and problem wording on strategy selection and downstream performance, and we validate these predictions in two new preregistered experiments. Overall, our work provides theoretical clarity and quantitative rigor to our understanding of intuitive judgment, and shows how such judgments can be understood as rational adaptations to the learning environment.
As large language models become primary communication partners, the stylistic and communicative character of their output—specifically the degree to which it relies on intellectual, affective, or action-oriented language—shapes how users interpret, and act on what they read. Yet this property is rarely measured or controlled directly. We introduce the Intellect-Emotion-Action Profile (IEAP), a purpose-built lexical framework featuring an inductively constructed dictionary from AI-generated text. IEAP decomposes any response into the proportional usage of intellectual, affective, and action words. Using a single automated harvester, we elicited and scored responses from four contemporary architectures (Claude, ChatGPT, Grok, and Gemini) across five question domains under four instructional-mode directives that span a cold-to-fire register (n = 30 per architecture per domain). Three primary findings emerge. First, when no directive is applied, native communicative profiles are dominated by the underlying question, with architecture playing a secondary, architecture-dependent role; thus, the prompt establishes the baseline for the response. Second, an explicit mode directive displaces this profile substantially from its baseline. Specifically, the directives produced large, monotonic shifts in affective word usage, exceeding 50 percentage points in the most responsive domain with no reversals. Third, response depth alters length and conceptual breadth while leaving lexical composition essentially unchanged, establishing register and depth as orthogonal controls. These three dimensions correlated with human-rated NRC lexicons in the expected directions, supporting their validity. Within-domain robustness checks reproduce both the baseline profiles and the titration trajectories across multiple question framings. Because a measurable communicative property can be monitored and audited, these results position IEAP as a practical foundation for studying, comparing, and ultimately governing how AI systems communicate.
William C. Kouns· Frontiers in Artificial Inte...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
