03 · What You Need to Know
How AI Can Favor Dominant Theories and Influence Scientific Inquiry
What Makes a Scientific Theory Dominant?
A theory or framework may become dominant when it is widely used, frequently cited, institutionally established, or influential within a discipline.
Dominance may reflect substantial empirical support, explanatory usefulness, historical development, or the availability of established measurement instruments.
It may also reflect scholarly networks, publication practices, disciplinary conventions, and unequal visibility among research communities.
These explanations are not mutually exclusive.
For example, the Technology Acceptance Model has been extensively applied in information systems and educational technology research. Its prominence may make it an obvious candidate when investigating perceived usefulness and intention to use technology.
However, its relevance becomes less straightforward when the research question concerns institutional power, unequal access, or the social consequences of technology use.
Researchers should distinguish theoretical prominence from suitability for a specific explanatory task.
AI Learns From Unevenly Represented Scholarly Material
Generative AI systems learn patterns from large collections of information, while some research assistants also retrieve material from external scholarly sources.
These resources do not necessarily represent every discipline, language, country, or theoretical tradition equally.
Prominent theories may appear repeatedly in textbooks, journal articles, reviews, and instructional materials.
A model may therefore associate particular research topics with familiar frameworks.
For example, a request concerning technology adoption may readily produce recommendations involving perceived usefulness or behavioral intention.
Such recommendations may be relevant, but their frequency does not establish that other perspectives were adequately considered.
The broader issue of bias in AI training data is relevant when uneven scholarly representation influences generated recommendations.
AI May Confuse Familiarity With Explanatory Adequacy
When researchers request a theoretical framework, AI may provide one that is strongly associated with the topic in available academic material.
However, a theory's familiarity does not establish that its assumptions match the phenomenon being studied.
For example, a framework focused on individual technology acceptance may be useful for examining students' intention to use generative AI.
It may be less sufficient for explaining how institutional restrictions, assessment policies, or unequal access shape actual use.
Researchers should examine what the theory explains and what remains outside its intended scope.
The appropriate response is not necessarily to reject the familiar framework. It may be to refine the research question, justify a complementary perspective, or select a different theoretical approach.
Dominant Theories Can Shape Research Questions Before Data Collection
Theoretical frameworks influence which variables and relationships researchers consider important.
If AI repeatedly recommends a particular theory, researchers may begin designing their study around its constructs.
For example, an AI adoption study may focus on perceived usefulness, perceived ease of use, and behavioral intention because these constructs are familiar and supported by established instruments.
However, the resulting design may provide limited information about institutional governance, academic integrity, or structural inequalities unless these dimensions are incorporated appropriately.
The omission may reflect the chosen research scope rather than an inherent flaw in the theory.
Researchers should ensure that the theoretical framework follows the research problem rather than allowing AI recommendations to determine which problems are worth investigating.
AI Can Reinforce Theoretical Path Dependence
Path dependence occurs when earlier decisions influence later choices, making established approaches more likely to be repeated.
In research, frequently used theories may become increasingly attractive because they offer established instruments, familiar terminology, and recognizable publication precedents.
AI can potentially reinforce this pattern by repeatedly recommending frameworks that are already prominent.
Researchers may then adopt those frameworks, produce additional publications using them, and further increase their visibility.
This is a plausible mechanism of theoretical reinforcement, but demonstrating actual amplification requires evidence about recommendation patterns and subsequent research behavior.
The possibility of AI amplifying existing scientific research biases should therefore be examined rather than assumed.
Mainstream Perspectives Are Not Necessarily Incorrect
Criticism of theoretical dominance should not be confused with the claim that mainstream science is inherently unreliable.
Some theories become influential because they explain important phenomena and have been examined through extensive empirical research.
Researchers should not treat novelty or marginality as evidence of superior scientific quality.
An alternative perspective may offer important insights while also having conceptual weaknesses or limited empirical support.
The appropriate evaluation concerns explanatory relevance, evidence, assumptions, and the research question.
AI-assisted theoretical exploration should broaden consideration where justified without creating false equivalence between well-supported explanations and unsupported claims.
AI May Underrepresent Critical and Alternative Perspectives
Some research questions benefit from perspectives that challenge established assumptions.
For example, technology acceptance frameworks may help explain adoption intentions, while critical perspectives may examine institutional power, surveillance, access inequalities, or the distribution of technological benefits.
These perspectives address different dimensions of technology use and are not necessarily competing explanations of the same outcome.
An AI system that recommends only adoption-oriented frameworks may provide an incomplete set of options for a study concerned with broader social consequences.
Researchers should examine whether alternative theoretical traditions are relevant to the specific phenomenon.
The objective is conceptual adequacy rather than maximizing the number of theories included.
English-Language Scholarship May Receive Greater Visibility
Many internationally visible research publications are written in English.
AI systems may therefore have greater exposure to theoretical terminology and scholarly discussions represented in English-language sources.
Conceptual approaches developed or discussed primarily in other languages may be less visible.
For example, research involving locally situated educational experiences may benefit from concepts that are not easily represented through familiar English-language frameworks.
Researchers should consider whether English-language conceptual assumptions influence theoretical recommendations.
However, language origin alone does not determine whether a theory is appropriate. Conceptual fit should be evaluated using the available evidence.
AI Can Overlook Local and Indigenous Knowledge
Some research questions concern practices, relationships, and experiences grounded in particular communities.
AI-generated theoretical recommendations may prioritize internationally familiar categories while overlooking locally developed concepts or Indigenous knowledge traditions.
This can affect how research problems are framed and which explanations receive attention.
For example, a study of educational participation might interpret interpersonal responsibilities solely through individual motivation constructs.
Such an interpretation could be incomplete when participants describe obligations and relationships that require a different conceptual account.
Researchers should examine whether AI makes marginalized or locally specific knowledge less visible.
Local knowledge should be engaged critically and contextually rather than treated as a decorative addition to an otherwise predetermined framework.
AI May Present Theoretical Consensus More Strongly Than the Literature Supports
Generative AI often produces coherent summaries of complex scholarly discussions.
However, a field may contain disagreements concerning theoretical assumptions, measurement, or interpretation.
An AI-generated synthesis may describe one dominant framework as though it represents an uncontested consensus.
For example, a model may state that a particular theory is universally accepted for explaining technology adoption.
Such a claim would require verification against the relevant literature.
Researchers should distinguish widespread use from universal acceptance and examine whether significant criticisms or alternatives exist.
Scientific consensus should be established through evidence, not inferred from the confidence of an AI-generated explanation.
AI Can Favor Theories With Easily Available Instruments
Some frameworks have established questionnaires, scales, and analytical conventions.
These resources can make a theory attractive to researchers because they reduce the work required to operationalize constructs.
AI may recommend such frameworks because their instruments and terminology appear frequently in scholarly material.
However, instrument availability is not sufficient justification for theoretical selection.
A validated instrument may be unsuitable for a different population, language, or research context.
Researchers should evaluate construct validity, conceptual relevance, and measurement equivalence before adopting a framework.
Convenience should not determine what the research claims to explain.
Theoretical Recommendations Depend on Prompt Framing
The instructions given to AI influence which frameworks it proposes.
A prompt requesting established theories of technology adoption is likely to produce different recommendations from one asking for perspectives on institutional governance and educational inequality.
These differences may reflect the researcher's framing rather than systematic model bias.
Researchers should examine whether their prompts prematurely narrow the conceptual options.
For exploratory theory selection, it may be useful to request frameworks with different explanatory assumptions and then evaluate their suitability independently.
AI-generated alternatives should be treated as candidates rather than automatically accepted as credible theories.
How Can Researchers Compare Theoretical Frameworks Systematically?
Researchers should establish criteria derived from the research question before selecting a theory.
Relevant considerations include the phenomenon explained, underlying assumptions, unit of analysis, available empirical support, conceptual limitations, and suitability for the study population.
| Criterion |
Question to examine |
| Explanatory target |
What phenomenon is the theory intended to explain? |
| Unit of analysis |
Does it focus on individuals, institutions, interactions, or broader systems? |
| Core assumptions |
What conditions must hold for the theory to be applicable? |
| Empirical support |
What evidence supports the theory in relevant contexts? |
| Conceptual coverage |
Which important aspects of the research problem are included or excluded? |
| Measurement |
Can the proposed constructs be operationalized appropriately? |
| Contextual suitability |
Does the theory adequately represent the population and setting? |
| Alternatives |
Would another perspective explain relevant findings more effectively? |
AI can help organize this comparison, but the evaluations should be grounded in original theoretical and empirical publications.
Researchers Should Verify That Recommended Theories Actually Exist
Generative AI can produce plausible names for theories, frameworks, or scholarly models that do not correspond to established publications.
It may also misattribute theoretical concepts to particular authors or combine elements from different frameworks.
Researchers should verify the original source of each recommended theory.
Where possible, they should consult foundational publications and relevant subsequent critiques or developments.
A model's description of a theory should not substitute for examining how its authors defined the constructs and assumptions.
Bibliographic verification is especially important when AI recommends unfamiliar or supposedly emerging frameworks.
Combining Theories Is Not Automatically Better
Researchers sometimes respond to a narrow theoretical recommendation by combining several frameworks.
However, adding theories does not necessarily improve explanatory quality.
Frameworks may operate at different levels, rely on incompatible assumptions, or introduce overlapping constructs.
For example, combining an individual technology acceptance model with an institutional theory may be useful when the research question genuinely concerns both levels.
But the researcher must explain how the frameworks relate and what each contributes.
An unnecessarily complex theoretical framework can make the research less coherent rather than more comprehensive.
The objective is a justified conceptual structure, not the largest possible collection of theories.
AI Can Help Identify Theoretical Blind Spots
Although AI may favor familiar perspectives, it can also be used to question a proposed framework.
Researchers may ask which aspects of the research problem remain unexplained, which assumptions require evidence, or which alternative traditions address different dimensions of the phenomenon.
For example, a study initially framed around students' intention to use AI might also consider whether institutional rules and assessment design influence actual use.
These questions can support theoretical reflection.
However, AI-generated critiques must be evaluated against the relevant literature and the researcher's own methodological reasoning.
The value of AI lies in generating possibilities for examination, not in independently determining the correct theoretical framework.
Dominant Theory Bias Can Affect Literature Reviews and Research Gaps
If AI repeatedly recommends studies associated with one theoretical tradition, the researcher may develop an incomplete picture of the field.
For example, a literature review on AI in higher education may become dominated by adoption and acceptance studies while overlooking research concerning institutional governance or educational inequality.
The resulting research gap may reflect the restricted literature sample rather than the actual state of scholarship.
Researchers should evaluate bias in AI literature recommendations and verify proposed gaps through appropriate independent searches.
Theoretical diversity should be considered in relation to the research question rather than pursued as an end in itself.
How Can Researchers Test Whether AI Favors Dominant Theories?
A researcher may evaluate AI-generated theoretical recommendations using a defined set of research scenarios.
For example, prompts can describe different educational technology questions and request frameworks without naming preferred theories.
The researcher can record which theories are recommended, how frequently they appear, and whether relevant alternatives are omitted.
Independent disciplinary reviewers may assess the suitability of the recommendations against a reference set developed through appropriate literature searches.
Repeated testing can help distinguish occasional suggestions from recurring patterns.
However, the frequency of a recommendation does not independently establish bias. Researchers must examine whether the recommended theory is appropriate and whether omitted alternatives are relevant.
Theoretical Selection Should Remain a Researcher Decision
The choice of theory shapes research questions, constructs, instruments, analyses, and interpretations.
Researchers should therefore be able to explain why the selected framework is appropriate and what its limitations are.
AI-generated recommendations may assist with initial exploration or comparison, but they should not replace engagement with foundational literature and relevant critiques.
Where AI materially contributes to theoretical development, researchers should document its role according to applicable institutional and publication requirements.
Human researchers remain accountable for the conceptual coherence and scientific justification of the study.
