UNDERSTANDING NEURO-ARTIFICIAL INTELLIGENCE: EFFECTS ON GHANAIAN STUDENTS’ LEARNING, COGNITION AND ACADEMIC DEVELOPMENT
Dr. Derrick Kwaku Antwi, Ph.D.
Computer Science Department
Ghana Communication Technology University, Accra, Ghana
Email: dkantwi@gctu.edu.gh
Abstract
The rapid development of Artificial Intelligence (AI), particularly generative artificial intelligence (GenAI), is transforming how students search for information, communicate, solve problems, write, study and interact with educational technologies. This development raises an important question: does increasing interaction with intelligent machines strengthen students’ cognitive abilities or encourage the outsourcing of cognitive processes to machines? This article introduces the concept of Neuro-Artificial Intelligence (Neuro-AI) as a framework for examining the relationship between intelligent computational systems and human cognitive processes in educational environments. The study focuses specifically on Ghanaian students and considers critical thinking, creativity, memory, problem-solving, metacognition, learning autonomy and academic performance.
Also read: Front Pages: Top newspaper headlines for Tuesday, 6th October 2026
The article proposes a mixed-methods research design involving students in Ghanaian secondary and tertiary institutions. A conceptual model is developed in which AI exposure and AI interaction influence cognitive processing and learning behaviour, while AI literacy, critical-thinking ability, metacognitive regulation, teacher guidance and digital access moderate these relationships. An algebraic Neuro-AI Learning Model is proposed to estimate the relationship between AI interaction and student learning outcomes. The paper argues that AI should not be conceptualized simply as either beneficial or harmful. Rather, its effects depend on the manner, intensity, purpose and pedagogical context of use. Recent systematic reviews indicate that generative AI can function as a cognitive amplifier when embedded in structured learning but can also become a cognitive substitute under unguided use. A Ghana-specific research agenda is therefore required to understand the educational and cognitive consequences of AI adoption.
The article concludes that Ghanaian educational institutions should move beyond debates about whether students should use AI and instead develop frameworks for how students should think with AI without allowing AI to replace thinking. The proposed Neuro-AI framework provides a foundation for empirical investigation, curriculum development, AI literacy, assessment reform and responsible educational AI policy in Ghana.
Keywords: Neuro-Artificial Intelligence, Artificial Intelligence, Generative AI, Ghanaian Students, Cognitive Development, Critical Thinking, Cognitive Offloading, AI Literacy, Learning, Higher Education
1. Introduction
Artificial Intelligence has moved from being primarily a computational research field to becoming an increasingly visible component of everyday human activity. Educational environments are particularly affected because students can now interact directly with large language models, intelligent tutoring systems, automated writing assistants, recommendation systems, adaptive learning platforms and multimodal AI systems.
Generative AI systems such as ChatGPT and comparable platforms can produce explanations, summaries, essays, computer programs, mathematical solutions, study questions and alternative interpretations within seconds. This capability creates opportunities for personalized learning and academic support. At the same time, it raises concerns about whether students may increasingly delegate activities that traditionally required memory, analysis, reasoning, creativity and independent problem-solving.
UNESCO’s guidance on generative AI in education emphasizes a human-centred approach, including ethical, safe, equitable and meaningful use, while highlighting concerns relating to privacy, regulation, validation and pedagogical design.
The central concern of this article is therefore not simply whether AI improves academic performance. A student may obtain a better immediate answer from AI without necessarily developing the cognitive capacity required to independently produce or evaluate that answer. Consequently, educational research needs to distinguish performance assistance from cognitive development.
Recent systematic evidence supports this distinction. A 2026 review of 89 peer-reviewed studies found that generative AI’s effects on higher-order cognitive skills were neither uniformly positive nor negative. Positive effects occurred under some conditions, while over-reliance, reduced analytical autonomy and cognitive offloading were identified as important risks. The review proposed a distinction between AI functioning as a cognitive amplifier and AI functioning as a cognitive substitute.
This issue is particularly important for Ghana. Ghanaian students increasingly operate within digitally mediated learning environments, but access to technology, digital literacy, instructional quality, infrastructure and institutional AI policies vary considerably. A 2026 study on AI literacy and education policy in Ghana argues that international AI-literacy frameworks cannot simply be transferred into Ghana without attention to local governance, infrastructure, learner agency and data sovereignty.
Ghanaian research has already begun examining the cognitive consequences of generative AI. Essel and colleagues investigated ChatGPT’s effects on critical, creative and reflective thinking among university students in Ghana, demonstrating that the Ghanaian context is relevant to the broader international debate about AI and cognition.
This article extends that research direction by proposing a Neuro-Artificial Intelligence framework for examining how AI-mediated learning may interact with human cognitive processes.
2. Background to Neuro-Artificial Intelligence
2.1 Definition
Neuro-Artificial Intelligence may be broadly conceptualized as an interdisciplinary research area concerned with the relationship between:
1. human neurological and cognitive processes;
2. artificial intelligence systems;
3. human-computer interaction;
4. learning and decision-making; and
5. computational augmentation of human cognition.
For educational research, Neuro-AI can be operationalized as:
The systematic study of how interaction between human learners and intelligent computational systems influences cognitive processing, learning behaviour, decision-making and academic development.
This definition is deliberately broader than brain-computer interfaces. The present study is primarily concerned with cognitive interaction between students and AI, rather than claiming that ordinary generative AI systems directly measure or alter brain activity.
2.2 Neuro-AI and Education
Traditional educational technology research often evaluates whether a technology improves achievement, engagement or satisfaction. A Neuro-AI approach asks an additional question:
What happens to the student’s thinking while the student is using the technology?
For example, consider a student asked to solve a programming problem.
Traditional learning process
Problem → Student recalls knowledge → Student analyses → Student develops solution → Student evaluates → Answer
AI-assisted process
Problem → Student asks AI → AI generates solution → Student accepts/analyses → Answer
The second process may be faster. However, if the student merely copies the AI-generated solution, the student may achieve task completion without exercising the full reasoning process.
Conversely, if the student asks AI for alternative solutions, challenges the output, verifies sources, identifies errors and develops an independent explanation, AI may function as a cognitive scaffold.
Therefore:
[
AI\ Assistance \neq Cognitive\ Development
]
and:
[
AI\ Assistance + Cognitive\ Engagement \rightarrow Potential\ Cognitive\ Development
]
3. Statement of the Problem
The adoption of generative AI in education has developed faster than many educational systems have been able to develop appropriate pedagogical and assessment frameworks.
Students can now obtain instant assistance with assignments, examinations, programming, research, writing and information retrieval. This creates an educational paradox.
On one side, AI can:
– reduce barriers to information;
– provide immediate feedback;
– explain difficult concepts;
– generate examples;
– support personalized learning;
– assist students with disabilities;
– provide language support; and
– increase access to learning resources.
On the other side, inappropriate use can contribute to:
– cognitive offloading;
– reduced independent reasoning;
– over-reliance;
– shallow learning;
– uncritical acceptance of information;
– reduced originality;
– academic integrity problems; and
– weakening of learner autonomy.
A 2026 systematic review reported over-reliance as the leading cognitive risk among the reviewed studies, followed by reduced analytical autonomy and cognitive offloading.
The problem is therefore not simply the presence of AI but the relationship between the learner and AI.
For Ghanaian education, there is an additional problem: the evidence base remains comparatively limited when compared with research from larger international educational systems. There is a need for context-specific research that considers Ghana’s educational structure, technological infrastructure, student characteristics, teacher roles, digital inequality and institutional policies.
4. Purpose of the Study
The purpose of this study is to develop and empirically test a Neuro-Artificial Intelligence framework for understanding how interaction with AI affects Ghanaian students’ cognitive and academic development.
Specifically, the study seeks to:
1. examine the level of AI use among Ghanaian students;
2. determine how students use generative AI for learning;
3. investigate the relationship between AI use and critical thinking;
4. examine the relationship between AI use and creative thinking;
5. investigate AI’s relationship with problem-solving;
6. examine the relationship between AI use and learning autonomy;
7. investigate cognitive offloading and AI dependence;
8. determine whether AI literacy moderates the relationship between AI use and learning outcomes;
9. examine the role of teacher guidance;
10. develop a Ghana-specific Neuro-AI educational framework.
5. Research Questions
Main Research Question
How does interaction with Artificial Intelligence influence the cognitive and academic development of Ghanaian students?
Specific Research Questions
RQ1: What is the level and pattern of AI use among Ghanaian students?
RQ2: How does AI interaction influence students’ critical-thinking abilities?
RQ3: How does AI interaction influence students’ creativity?
RQ4: What relationship exists between AI use and students’ problem-solving abilities?
RQ5: Does excessive AI dependence contribute to cognitive offloading?
RQ6: Does AI literacy influence the relationship between AI use and academic performance?
RQ7: Does teacher guidance moderate the effects of AI on student cognition?
RQ8: What Neuro-AI framework can support responsible AI-enabled education in Ghana?
6. Research Hypotheses
The proposed study will test the following hypotheses.
H1
AI interaction has a significant relationship with students’ academic learning outcomes.
H2
Structured AI use has a positive relationship with students’ critical-thinking ability.
H3
Unguided AI dependence has a negative relationship with students’ independent problem-solving ability.
H4
AI literacy significantly moderates the relationship between AI use and cognitive development.
H5
Metacognitive regulation significantly moderates the relationship between AI use and learning outcomes.
H6
Teacher guidance significantly moderates the relationship between AI use and students’ cognitive development.
H7
Cognitive offloading mediates the relationship between excessive AI dependence and independent reasoning.
H8
Students who use AI as a cognitive scaffold demonstrate stronger learning autonomy than students who primarily use AI as a replacement for independent work.
7. Literature Review
7.1 Artificial Intelligence in Education
Artificial Intelligence in Education involves the application of computational intelligence to teaching, learning, assessment, feedback, administration and educational research.
Examples include:
– intelligent tutoring systems;
– automated assessment;
– adaptive learning;
– learning analytics;
– conversational AI;
– generative AI;
– predictive analytics;
– recommendation systems; and
– educational robotics.
Generative AI has expanded this field because students can directly interact with systems capable of generating natural-language explanations and other forms of content.
UNESCO recommends that educational AI adoption should remain human-centred and should address safety, equity, privacy, ethics and meaningful educational use.
7.2 AI and Critical Thinking
Critical thinking involves the capacity to analyse information, evaluate evidence, identify assumptions, compare alternatives and make reasoned judgments.
Generative AI may support critical thinking when students are required to evaluate AI-generated answers rather than merely accept them.
For example:
AI statement → Student verifies → Student compares evidence → Student identifies errors → Student develops conclusion
This process can activate analytical reasoning.
However:
AI statement → Student accepts → Student copies
may reduce opportunities for independent evaluation.
Recent systematic research indicates that structured and guided use appears more beneficial for critical thinking than passive or unguided use.
7.3 AI and Creativity
AI can generate large numbers of ideas rapidly. This can support brainstorming and divergent thinking.
A student writing a research proposal, for example, could request:
– alternative research questions;
– competing explanations;
– possible methodologies;
– conceptual models; and
– counterarguments.
However, creativity requires more than producing many ideas. It involves originality, appropriateness, evaluation and transformation.
Consequently, AI may function as:
[
AI \rightarrow Idea\ Generator
]
rather than:
[
AI \rightarrow Human\ Creativity\ Replacement
]
The student’s own evaluation and transformation of AI-generated ideas remain essential.
7.4 AI and Problem-Solving
Problem-solving requires understanding the problem, generating possible solutions, testing alternatives and evaluating outcomes.
AI can accelerate these activities. In engineering, mathematics, computer science and other disciplines, AI may provide possible solutions or simulations.
However, a student who cannot independently explain why an AI-generated solution works may have achieved task completion without acquiring transferable knowledge.
The relevant distinction is therefore:
[
AI\ Supported\ Problem\ Solving
]
versus
[
AI\ Substituted\ Problem\ Solving
]
The first maintains student agency; the second potentially reduces it.
8. Cognitive Offloading
Cognitive offloading refers to the delegation of mental tasks to external tools.
Students have always used external cognitive resources. Calculators, textbooks, search engines and note-taking systems can all reduce cognitive demands.
The important question is whether offloading facilitates learning or replaces the cognitive activity that learning is intended to develop.
The emerging GenAI literature identifies cognitive offloading as a significant issue. A 2026 systematic review found that AI’s effects depend strongly on scaffolding, learner prior knowledge, dosage and task design.
The proposed model therefore distinguishes:
Productive offloading
[
Offloading \rightarrow Reduced\ unnecessary\ load \rightarrow More\ cognitive\ resources\ for\ learning
]
from:
Substitutive offloading
[
Offloading \rightarrow Reduced\ cognitive\ engagement \rightarrow Reduced\ independent\ learning
]
9. Theoretical Framework
The study integrates four major theoretical perspectives.
9.1 Cognitive Load Theory
Cognitive Load Theory proposes that working memory has limited capacity. Educational technologies can potentially reduce unnecessary cognitive demands.
However, reducing cognitive load is not automatically beneficial if the removed activity is itself necessary for learning.
9.2 Constructivist Learning Theory
Constructivism views learners as active participants in knowledge construction.
Under this perspective, AI should function as a resource that supports learners in constructing knowledge rather than replacing the learning process.
9.3 Self-Regulated Learning Theory
Self-regulated learners plan, monitor and evaluate their learning.
AI literacy therefore needs to include metacognitive skills:
Plan → Ask → Evaluate → Verify → Reflect → Apply
9.4 Cognitive Offloading Theory
Cognitive offloading explains why humans transfer mental tasks to external systems.
The Neuro-AI model extends this idea by asking whether AI-mediated offloading results in:
augmentation or substitution.
10. Proposed Neuro-AI Conceptual Framework
The proposed framework is:
ARTIFICIAL INTELLIGENCE
│
▼
AI INTERACTION
│
┌────────────┼────────────┐
▼ ▼ ▼
AI SUPPORT AI DIALOGUE AI DEPENDENCE
│ │ │
▼ ▼ ▼
COGNITIVE COGNITIVE COGNITIVE
SCAFFOLD ENGAGEMENT OFFLOADING
│ │ │
└────────────┼────────────┘
▼
COGNITIVE PROCESSING
│
┌────────────┼────────────┐
▼ ▼ ▼
CRITICAL CREATIVE PROBLEM
THINKING THINKING SOLVING
│ │ │
└────────────┼────────────┘
▼
LEARNING OUTCOME
│
▼
ACADEMIC DEVELOPMENT
The relationship is moderated by:
– AI literacy;
– teacher guidance;
– metacognitive regulation;
– digital access;
– prior knowledge;
– discipline;
– task complexity; and
– institutional AI policy.
11. Neuro-AI Learning Equation
A conceptual mathematical model can be expressed as:
[
L = \beta_0 + \beta_1AI + \beta_2AL + \beta_3CT + \beta_4MR + \beta_5TG – \beta_6AD + \epsilon
]
Where:
– L = learning outcome;
– AI = intensity/quality of AI interaction;
– AL = AI literacy;
– CT = critical-thinking capacity;
– MR = metacognitive regulation;
– TG = teacher guidance;
– AD = AI dependence;
– \beta_0 = intercept;
– \beta_1…\beta_6 = regression coefficients;
– \epsilon = unexplained error.
A broader Neuro-AI Cognitive Development Index can be represented as:
[
NCDI = w_1CT+w_2CR+w_3PS+w_4LA+w_5MR-w_6CO
]
Where:
– CT = critical thinking;
– CR = creativity;
– PS = problem-solving;
– LA = learning autonomy;
– MR = metacognitive regulation;
– CO = cognitive offloading;
– w = weighting coefficients.
The model assumes that AI use can have both positive and negative effects.
Therefore:
[
AI_{effect}=AI_{augmentation}-AI_{substitution}
]
This represents the central proposition of the paper:
«The educational value of AI depends not only on how much AI a student uses, but on what cognitive functions the student retains while using it.»
12. Methodology
12.1 Research Design
A mixed-methods research design is proposed.
The quantitative component will examine statistical relationships among AI use, cognitive development and academic outcomes.
The qualitative component will explore students’ experiences, perceptions and strategies when interacting with AI.
12.2 Study Population
The population will consist of students enrolled in selected Ghanaian educational institutions.
The study may include:
– Senior High School students;
– undergraduate students;
– postgraduate students; and
– selected professional/technical students.
For the initial empirical phase, the recommended primary population is tertiary students in Ghana, because they are likely to have substantial exposure to generative AI for academic work.
12.3 Sample Size
A proposed sample of approximately 400–600 students would provide a useful starting point for quantitative analysis, subject to formal power analysis and institutional access.
Sampling may use a multistage approach:
1. institution selection;
2. programme/discipline selection;
3. class-level selection;
4. student selection.
The final sample size should be determined using statistical power analysis rather than convenience alone.
13. Data Collection Instrument
A structured questionnaire can contain six sections.
Section A: Demographics
– age group;
– sex/gender;
– academic level;
– programme;
– institution;
– location;
– access to smartphone/computer;
– internet access.
Section B: AI Usage
Students indicate frequency of:
– ChatGPT;
– Gemini;
– Copilot;
– AI writing tools;
– AI search systems;
– AI programming assistants;
– AI tutoring systems.
Section C: Cognitive Engagement
Example statements:
1. I evaluate AI-generated answers before accepting them.
2. I compare AI responses with other sources.
3. I ask AI to explain why an answer is correct.
4. I identify errors in AI-generated responses.
5. I develop my own solution before consulting AI.
Section D: AI Dependence
Example statements:
1. I find it difficult to complete assignments without AI.
2. I regularly copy AI-generated responses into academic work.
3. I ask AI to solve problems before attempting them myself.
4. I depend on AI to generate ideas for most assignments.
Section E: Cognitive Development
Measures may include:
– critical thinking;
– creativity;
– problem-solving;
– learning autonomy;
– metacognition.
Section F: Academic Outcomes
Possible measures include:
– self-reported academic performance;
– assignment performance;
– course grades where ethically and institutionally permitted;
– perceived learning;
– retention;
– confidence.
A five-point Likert scale can be used:
1 = Strongly Disagree
2 = Disagree
3 = Neutral
4 = Agree
5 = Strongly Agree
14. Validity and Reliability
Content validity should be assessed by experts in:
– artificial intelligence;
– educational technology;
– psychology/cognitive science;
– computer science;
– educational research.
A pilot study involving approximately 30–50 students can be conducted before the main study.
Internal consistency may be assessed using Cronbach’s alpha:
[
\alpha = \frac{k}{k-1}
\left(1-\frac{\sum\sigma_i^2}{\sigma_T^2}\right)
]
where:
– k = number of items;
– \sigma_i^2 = variance of each item;
– \sigma_T^2 = total variance.
A coefficient of approximately .70 or above is commonly considered acceptable for exploratory research, although reliability should be interpreted alongside construct validity.
15. Data Analysis
The quantitative data can be analysed using:
– descriptive statistics;
– frequencies;
– percentages;
– means;
– standard deviations;
– Pearson correlation;
– multiple regression;
– mediation analysis;
– moderation analysis;
– structural equation modelling where sample size permits.
The basic regression model can be expressed as:
[
Y=\beta_0+\beta_1X_1+\beta_2X_2+\cdots+\beta_nX_n+\epsilon
]
The mediation hypothesis can examine:
[
AI\ Dependence \rightarrow Cognitive\ Offloading \rightarrow Independent\ Reasoning
]
Moderation can examine:
[
AI\ Use \times AI\ Literacy \rightarrow Cognitive\ Development
]
Qualitative responses can be analysed using thematic analysis.
16. Proposed Analytical Model
Variable| Role| Example indicators
AI interaction| Independent| Frequency, duration, task type
AI literacy| Moderator| Understanding AI limitations
Teacher guidance| Moderator| Instruction and supervision
Metacognition| Moderator| Planning, monitoring, reflection
Cognitive offloading| Mediator| Delegation of reasoning
Critical thinking| Outcome| Evaluation and analysis
Creativity| Outcome| Originality and idea generation
Problem-solving| Outcome| Independent solution development
Learning autonomy| Outcome| Independent study
Academic performance| Outcome| Grades/perceived achievement
17. Expected Findings
Because this manuscript proposes an empirical study rather than reporting completed data, the following are expected relationships, not claimed results.
It is anticipated that moderate and structured AI use will be positively associated with learning outcomes when students actively evaluate and interrogate AI outputs.
It is also anticipated that excessive dependence may be associated with greater cognitive offloading and reduced independent reasoning.
The expected pattern can be represented as an inverted or conditional relationship:
Learning
Outcome
^
| ______
| __/
| __/
| __/
|______/____________________> AI Use
Low Optimal Excessive
The implication is not that more AI is necessarily better. Rather, there may be an optimal zone of AI-supported learning in which AI assists cognition without replacing the student’s cognitive work.
Recent systematic evidence is consistent with this conditional interpretation. A 2026 review found that GenAI’s effects varied according to instructional strategy and identified a dual role in which AI can act as an amplifier under structured conditions and a substitute under unguided conditions.
18. Discussion
The central argument of this paper is that the relationship between students and AI should be treated as a cognitive relationship rather than simply a technological one.
When students interact with AI, at least three processes may occur.
Process 1: Cognitive Extension
AI provides information that expands the student’s cognitive resources.
[
Human\ cognition + AI\ support \rightarrow Extended\ capability
]
Process 2: Cognitive Collaboration
The student and AI engage in an iterative process.
[
Student \rightarrow AI \rightarrow Student \rightarrow Verification \rightarrow Refinement
]
This may be particularly useful for inquiry-based learning.
Process 3: Cognitive Substitution
The student transfers the intellectual task to AI.
[
Problem \rightarrow AI \rightarrow Answer
]
The third process creates the greatest concern because the student may receive an answer without performing the cognitive processes necessary to learn.
This distinction is increasingly supported by the research literature. A 2026 systematic review of higher-order cognitive outcomes found that positive outcomes were conditional and that over-reliance, reduced analytical autonomy and cognitive offloading were important risks.
Similarly, a 2026 review of cognitive load and generative AI concluded that benefits depend on factors including scaffolding, prior knowledge, dosage and task design.
Therefore, the question for Ghanaian education should not simply be:
«“Should students use AI?”»
The more productive question is:
«“How can Ghanaian students use AI while preserving and strengthening human cognitive capability?”»
19. Neuro-AI and Ghanaian Education
The Ghanaian context requires particular attention to infrastructure and educational inequality.
Students may differ in:
– internet access;
– device availability;
– AI subscriptions;
– digital literacy;
– English-language proficiency;
– teacher support;
– institutional policies;
– exposure to AI training.
Consequently, AI may simultaneously reduce some educational inequalities while creating others.
For example, students with reliable connectivity and advanced AI literacy may gain substantially more from generative AI than students who have limited digital access.
The 2026 Ghana-focused AI-literacy literature highlights the importance of local governance, learner agency, data sovereignty and the limitations of directly importing global AI-literacy models into Ghana.
A Ghanaian Neuro-AI framework should therefore include:
1. AI Literacy
Students need to understand:
– what AI can do;
– what AI cannot do;
– hallucination;
– bias;
– source verification;
– privacy;
– responsible prompting;
– academic integrity.
2. Cognitive Literacy
Students should understand:
– memory;
– attention;
– reasoning;
– metacognition;
– critical thinking;
– cognitive offloading.
3. Ethical Literacy
Students need to understand:
– plagiarism;
– authorship;
– privacy;
– intellectual property;
– responsible AI use.
4. Teacher AI Competence
Teachers should learn how to redesign assignments so that students demonstrate their reasoning rather than merely submitting AI-generated products.
20. Implications for Teaching
Teachers should shift from assessing only the final product to assessing the learning process.
Instead of asking:
«“Submit a 2,000-word essay.”»
Teachers could ask students to submit:
1. initial research question;
2. independent argument;
3. AI interaction record;
4. verification of AI claims;
5. revised argument;
6. reflection on what AI contributed;
7. final academic product.
This produces a new assessment model:
[
Assessment = Product + Process + Reflection
]
This approach can make AI use visible while maintaining student accountability.
21. Proposed Neuro-AI Classroom Model
TEACHER
│
▼
LEARNING PROBLEM
│
▼
STUDENT THINKING
│
▼
AI INTERACTION
│
┌───────┴───────┐
▼ ▼
AI RESPONSE ALTERNATIVES
│ │
└───────┬───────┘
▼
STUDENT EVALUATION
│
▼
SOURCE VERIFICATION
│
▼
STUDENT REASONING
│
▼
REFLECTION
│
▼
LEARNING OUTCOME
The teacher remains an important component of the system.
AI should therefore be viewed as:
[
Teacher + Student + AI
]
rather than:
[
AI \rightarrow Student
]
22. Policy Implications
Ghanaian educational institutions should develop institutional AI policies addressing:
1. acceptable AI use;
2. prohibited AI use;
3. AI disclosure;
4. academic integrity;
5. student privacy;
6. assessment;
7. AI literacy;
8. staff development;
9. data governance;
10. accessibility.
Institutions should avoid policies based solely on prohibition because students may continue using AI outside institutional visibility.
A more sustainable model is:
[
Regulation + Education + Responsible\ Use
]
UNESCO similarly advocates coherent policy frameworks that promote ethical, safe, equitable and meaningful use of generative AI in education.
23. Ethical Considerations
The proposed study should receive appropriate institutional ethical approval before data collection.
Participants should receive information concerning:
– study purpose;
– voluntary participation;
– confidentiality;
– withdrawal rights;
– data storage;
– intended use of findings.
Students should not be required to disclose private AI conversations unless ethically approved and necessary for the research.
Where minors are included, appropriate parental/guardian consent and student assent procedures should be followed.
No claim should be made that ordinary interaction with ChatGPT or similar systems directly changes the physical structure or function of a student’s brain unless supported by appropriate neuroscientific evidence.
This distinction is particularly important because Neuro-AI in this paper is principally a cognitive and interdisciplinary framework, not a claim that generative AI itself is a neurotechnology.
24. Limitations
Several limitations should be recognized.
First, a cross-sectional study cannot establish long-term causality. Students who are already strong academically may use AI differently from students who struggle academically.
Second, self-reported AI usage may contain reporting bias.
Third, AI systems change rapidly, meaning that findings concerning a particular platform or model may become outdated.
Fourth, Ghana contains diverse educational, socioeconomic and linguistic contexts, so results from selected institutions may not represent all Ghanaian students.
Fifth, measuring cognition using questionnaires alone may be insufficient. Future studies should combine surveys with objective cognitive assessments, academic performance measures and, where ethically and technically justified, neuroscientific methods.
25. Future Research
Future research should develop a longitudinal Ghanaian Neuro-AI research programme.
Phase 1: Baseline
Measure:
– AI use;
– AI literacy;
– critical thinking;
– creativity;
– problem-solving;
– learning autonomy.
Phase 2: Intervention
Introduce structured AI-literacy training.
Phase 3: Experimental learning
Compare:
Group A: Traditional learning
Group B: Unguided AI learning
Group C: Guided Neuro-AI learning
Phase 4: Assessment
Compare:
– academic performance;
– critical thinking;
– creativity;
– problem-solving;
– retention;
– AI dependence.
Phase 5: Longitudinal study
Track students over one or more academic years.
This design would provide stronger evidence regarding whether AI functions as a cognitive amplifier or substitute.
26. Conclusion
Artificial Intelligence is changing the relationship between students, knowledge and educational institutions. The central educational challenge is no longer simply access to information. Students increasingly have access to systems capable of generating information, explanations and solutions instantly.
The more important challenge is therefore the preservation and development of human cognitive capacity.
This article proposes Neuro-Artificial Intelligence as a framework for examining this relationship among Ghanaian students. The framework conceptualizes AI interaction as potentially producing cognitive augmentation, cognitive collaboration or cognitive substitution.
The distinction is fundamental.
When students use AI to question, compare, verify, analyse, reflect and improve their own ideas, AI can function as a cognitive scaffold.
When students use AI to avoid thinking, reasoning and problem-solving, AI may become a cognitive substitute.
Therefore:
[
Responsible\ Neuro!-!AI =
AI\ Capability + Human\ Cognition + Critical\ Evaluation + Ethical\ Regulation
]
The future of AI-enabled education in Ghana should not be built around replacing teachers or replacing student thinking. It should be built around strengthening human intelligence through responsible interaction with artificial intelligence.
The ultimate objective should be:
«AI should help Ghanaian students think better, not think less.»
27. Recommendations
Recommendation 1: Establish AI Literacy Programmes
Universities and schools should introduce structured AI-literacy programmes covering AI capabilities, limitations, bias, verification, prompting, privacy and academic integrity.
Recommendation 2: Introduce Neuro-AI Literacy
Students should learn about cognitive offloading, metacognition, attention, memory and independent reasoning.
Recommendation 3: Redesign Assessments
Assessment should emphasize reasoning, oral defence, reflection, project development and process documentation.
Recommendation 4: Train Teachers
Teachers should receive professional development in AI-supported pedagogy and AI-resistant assessment design.
Recommendation 5: Develop Institutional AI Policies
Institutions should establish clear rules for acceptable and unacceptable AI use.
Recommendation 6: Promote Ghana-Specific Research
Research should examine Ghanaian students across different regions, institutions, disciplines and socioeconomic groups.
Recommendation 7: Protect Student Data
Educational AI systems should follow strong privacy and data-governance principles.
Recommendation 8: Encourage Human-AI Collaboration
Students should be trained to use AI as an intellectual partner rather than an unquestioned authority.
28. Proposed Contribution of the Study
The proposed study contributes three major ideas.
First: Neuro-AI Educational Perspective
It connects AI-in-education research with cognitive development.
Second: Ghanaian Context
It provides a framework specifically designed for investigating Ghanaian students rather than assuming that findings from other educational systems automatically apply to Ghana.
Third: Cognitive Amplifier–Substitute Model
It proposes that AI’s educational impact depends on whether the technology:
[
Amplifies\ human\ cognition
]
or
[
Substitutes\ for\ human\ cognition.
]
This distinction provides a foundation for future empirical research.
References
Anderson, L. W., & Krathwohl, D. R. (Eds.). (2001). A taxonomy for learning, teaching, and assessing: A revision of Bloom’s taxonomy of educational objectives. Longman.
Essel, H. B., Vlachopoulos, D., Essuman, A. B., & Amankwa, J. O. (2024). ChatGPT effects on cognitive skills of undergraduate students: Receiving instant responses from AI-based conversational large language models. Computers and Education: Artificial Intelligence, 6, 100198. https://doi.org/10.1016/j.caeai.2023.100198
Kasneci, E., Sessler, K., Küchemann, S., Bannert, M., Dementieva, D., Fischer, F., Gasser, U., Groh, G., Günnemann, S., Hüllermeier, E., Krusche, S., Kutyniok, G., Michaeli, T., Nerdel, C., Pfeiffer, F., Poquet, O., Sailer, M., Schmidt, A., Seidel, T., … Kasneci, G. (2023). ChatGPT for good? On opportunities and challenges of large language models for education. Learning and Individual Differences, 103, 102274. https://doi.org/10.1016/j.lindif.2023.102274
Lo, C. K. (2023). What is the impact of ChatGPT on education? A rapid review of the literature. Education Sciences, 13(4), 410.
Miao, F., & Holmes, W. (2023). Guidance for generative AI in education and research. UNESCO.
Nasr, N. R., Tu, C.-H., Sujo-Montes, L., Haniya, S., Yen, C.-J., et al. (2026). Generative artificial intelligence impact on critical thinking in higher education: A comprehensive bibliometric analysis and systematic review. Journal of Computers in Education.
Qian, W., Yang, F., Cao, Y., Yi, L., Gu, R., & Wang, Z. (2026). Generative AI and cognitive load in education: A systematic review of WoS/SSCI-indexed studies through the lens of cognitive load theory. Frontiers in Psychology, 17, 1921504.
Alubthane, F. O. (2026). Amplifier or substitute? A systematic review of generative AI’s impact on higher-order cognitive skills among university students. Frontiers in Psychology, 17, 1863931. https://doi.org/10.3389/fpsyg.2026.1863931
Suazo, I., Santiesteban Velázquez, M., Saracostti, M., & Chaple Gil, A. M. (2026). Between cognitive offloading and critical autonomy: A systematic review of the epistemic implications of generative AI in higher education. Frontiers in Education, 11, 1878665.
UNESCO. (2023). Guidance for generative AI in education and research. UNESCO.
Zawacki-Richter, O., Marín, V. I., Bond, M., & Gouverneur, F. (2019). Systematic review of research on artificial intelligence applications in higher education—Where are the educators? International Journal of Educational Technology in Higher Education, 16, 39.
Author Note
Dr. Derrick Kwaku Antwi, Ph.D. is a researcher and lecturer in the field of computer science and artificial intelligence. His research interests include Artificial Intelligence, Robotics, Internet of Things, digital education, intelligent systems and emerging technologies.
Correspondence: dkantwi@gctu.edu.gh