Generative AI as a Cognitive Amplifier or Substitute: A Computational Framework for Human Reasoning and Decision Intelligence
Year: 2026
Type: article
Abstract: Generative Artificial Intelligence (GenAI) is now part and parcel of learning, problem-solving, and decision-making, but the cognitive impact is still debated as it can complement and supplant some aspects of human cognition. The study investigated the cognitive amplification and cognitive substitution of GenAI using a computational model that incorporates the cognitive reliance of using AI, cognitive engagement, cognitive offloading, amplification and substitution. A cleaned sample of 394 respondents was used and a quantitative secondary data design was used. Relationships and distinguishable human-AI interaction profiles were identified using descriptive statistics, reliability analysis, correlation analysis, multiple regression, clustering and machine-learning classification analysis. Cognitive amplification had the highest overall mean with cognitive substitution and offloading also being significant. Cognitive offloading was the best predictor of both amplification and substitution, suggesting that less cognitive effort can help performance, and enhance reliance as well. Both human cognitive involvement and AI reliance were highly predictive of amplification, while human cognitive involvement was significantly predictive of substitution. Four clusters were identified: moderate AI users, high-substitution/high-offloading users, low-engagement users, and high-amplification/active-engagement users. The random forest method was highly successful in achieving predictive classification of these profiles. The results show that when using GenAI, it is not just a booster or a replacement, but the interplay between the extent of AI reliance, cognitive offloading, verification, and ongoing human involvement are what determines its impact.
Source: Computer and decision making.
Language: English
Subfield: Human Factors and Ergonomics
Field: Social Sciences
Domain: Social Sciences
Open Access status: hybrid
OpenAlex ID: W7225138803
