AI-INDUCED CONCEPTUAL DRIFT

The increasing use of generative artificial intelligence (AI) as an explanatory and educational partner raises important questions about how repeated interactions may influence human conceptual understanding. While existing research has examined automation bias, misinformation effects, the illusion of explanatory depth, and AI-related changes in cognitive performance, the gradual transformation of an individual's conceptual representations across repeated AI-assisted explanations requires more systematic investigation. This article develops a conceptual framework for AI-Induced Conceptual Drift (AICD), provisionally defined as a measurable change in how an individual represents, interprets, distinguishes, and applies a concept over time, where repeated exposure to AI-generated explanations contributes causally to that change. Drawing on conceptual change theory, cognitive psychology, metacognition, and human–AI interaction research, the article proposes the Conceptual Drift Pathways Model (CDPM). The model identifies six possible pathways of change: terminological assimilation, category-boundary modification, explanatory restructuring, inferential transfer, epistemic recalibration, and normative reframing. It distinguishes beneficial conceptual development from misleading or unjustified conceptual displacement and outlines the psychological mechanisms through which repeated explanations may influence both. A longitudinal experimental design is proposed to measure changes in concept definitions, classification decisions, causal explanations, confidence calibration, and independent transfer performance. The paper argues that evaluating AI-assisted learning requires attention not only to immediate task performance or subjective understanding but also to the stability, accuracy, flexibility, and independent usability of conceptual knowledge over time.

Authors

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-14
DOI
https://doi.org/10.5281/zenodo.24217521
Primary Topic
Intelligent Tutoring Systems and Adaptive Learning
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

AI-INDUCED CONCEPTUAL DRIFT

Roman Shaban
Zenodo (CERN European Organization for Nuclear Research)
Intelligent Tutoring Systems and Adaptive Learning
preprint

AI-INDUCED CONCEPTUAL DRIFT

Roman Shaban
preprint en

Abstract

The increasing use of generative artificial intelligence (AI) as an explanatory and educational partner raises important questions about how repeated interactions may influence human conceptual understanding. While existing research has examined automation bias, misinformation effects, the illusion of explanatory depth, and AI-related changes in cognitive performance, the gradual transformation of an individual's conceptual representations across repeated AI-assisted explanations requires more systematic investigation. This article develops a conceptual framework for AI-Induced Conceptual Drift (AICD), provisionally defined as a measurable change in how an individual represents, interprets, distinguishes, and applies a concept over time, where repeated exposure to AI-generated explanations contributes causally to that change. Drawing on conceptual change theory, cognitive psychology, metacognition, and human–AI interaction research, the article proposes the Conceptual Drift Pathways Model (CDPM). The model identifies six possible pathways of change: terminological assimilation, category-boundary modification, explanatory restructuring, inferential transfer, epistemic recalibration, and normative reframing. It distinguishes beneficial conceptual development from misleading or unjustified conceptual displacement and outlines the psychological mechanisms through which repeated explanations may influence both. A longitudinal experimental design is proposed to measure changes in concept definitions, classification decisions, causal explanations, confidence calibration, and independent transfer performance. The paper argues that evaluating AI-assisted learning requires attention not only to immediate task performance or subjective understanding but also to the stability, accuracy, flexibility, and independent usability of conceptual knowledge over time.

Zenodo (CERN European Organization for Nuclear Research)
Intelligent Tutoring Systems and Adaptive Learning
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon