Integrating Universal Generative AI Platforms in Educational Labs to Foster Critical Thinking and Digital Literacy

International Journal on Cybernetics & Informatics (IJCI) Vol.14, No.3, June 2025
Bibhu Dash et al: CSEAI, SCDD, CIOS - 2025
pp. 13-25, 2025. IJCI – 2025 DOI:10.5121/ijci.2025.140302
INTEGRATING UNIVERSAL GENERATIVE AI
PLATFORMS IN EDUCATIONAL LABS TO FOSTER
CRITICAL THINKING AND DIGITAL LITERACY
Vasiliy Znamenskiy, Rafael Niyazov, and Joel Hernandez
Science Department, Borough of Manhattan Community College, The City University of
New York, New York, USA
ABSTRACT
This study investigates the educational potential of generative artificial intelligence (GenAI) platforms
based on large language models (LLMs), such as ChatGPT, Claude, and Gemini, as tools for student-
centred learning. Recognizing the current limitations of GenAI, particularly its propensity for generating
inaccurate or misleading information—the paper proposes a novel instructional strategy: an
interdisciplinary laboratory designed to foster critical evaluation of GenAI-generated outputs.
In this pedagogical model, students engage with GenAI systems by posing questions or solving problems
drawn from topics they have already studied and understand. Equipped with correct answers, students are
positioned to assess the accuracy and relevance of AI-generated responses across multiple modalities,
including text, images, and video.
Students design challenging prompts and tasks that help compare the intellectual performance of various
GenAI. This approach was implemented for a specially designed lab session within a general astronomy
course for non-science majors. Multiple student groups completed the lab, demonstrating high levels of
engagement, initiative, and critical thinking. Findings suggest that such activities not only deepen students’
comprehension of the scientific content of learning courses but also cultivate essential skills in digital
literacy and critical interaction with AI technologies.
KEYWORDS
Generative Artificial Intelligence, GenAI, Large Language Models, LLMs, Critical Thinking, Digital
Literacy, AI in Education, ChatGPT, Claude, Gemini, AI-augmented Learning, Interdisciplinary Education
1. INTRODUCTION
The current phase in the evolution of artificial intelligence is marked by the rapid advancement
and widespread adoption of large language models (LLMs), such as ChatGPT (OpenAI), Gemini
(Google), and Claude (Anthropic) [1]. While earlier dialogue systems like ELIZA and ALICE
laid the foundational groundwork, recent progress in transformer architectures, deep learning, and
cloud-based computing [2] has resulted in a significant leap in the performance and generative
capabilities of modern systems. These advances have expanded the functionality of LLMs from
basic language comprehension to the generation of coherent text, executable code, images, and
video content.
Contemporary generative AI (GenAI) dialogue systems exhibit notable proficiency in processing
open-ended user inputs and generating diverse outputs, including written texts, multimedia
materials, and programming scripts. Their rapid integration into everyday life has sparked

14
widespread public debate, bringing AI-related discussions beyond technical and scientific
contexts [3] into educational, ethical, and sociopolitical domains.
In education, the rise of GenAI technologies has elicited mixed responses. On the one hand, these
tools promise to support personalized learning, enhance student motivation, and facilitate more
effective pedagogical interactions [4][5] through intelligent tutoring systems and tailored
educational chatbots. On the other hand, concerns are growing about the erosion of academic
integrity [5][6], as students increasingly rely on GenAI tools for completing assessments and
assignments without genuine engagement, potentially leading to superficial learning and
diminished cognitive development.
Developing competencies for the responsible and informed use of GenAI [2][7] is becoming
essential for 21st-century graduates. Within the broader agenda of digital transformation in
higher education [3], institutions bear a responsibility to promote the ethical deployment of AI
for societal benefit. This includes embedding critical digital literacy into curricula [3][6] and
offering systematic training for both students and faculty on the legal, ethical, and data security
dimensions of AI use. Such efforts support not only individual academic and professional growth
but also the formation of a more inclusive and equitable digital society.
Despite the potential of GenAI, intelligent educational assistants and domain-specific chatbots
[8][9] are not yet widely used in postsecondary teaching practices. However, general-purpose
LLMs are increasingly being adopted by students, often as tools for generating assignment
responses with minimal engagement in the learning process. This trend necessitates a critical
reassessment by educators of their pedagogical stance toward AI usage in academic work.
One response has been to limit digital access in classrooms through bans on smartphones or
restricted internet connectivity. An alternative and more constructive strategy is to incorporate
chatbot interactions into formal instruction [4]. In this model, educators serve as guides and
facilitators, helping students learn to engage with GenAI systems critically and effectively. Given
that LLMs, while powerful, are prone to producing plausible but erroneous or misleading outputs
[6][9] —sometimes in unexpectedly straightforward contexts—students must be equipped not
only with the ability to formulate effective prompts but also to critically evaluate AI-generated
information.
Consequently, there is an emerging need to design interdisciplinary educational tasks [7][9] that
simultaneously support subject-matter learning, cultivate critical thinking, and foster responsible
AI engagement skills.
While banning or limiting access to AI may seem like a straightforward solution, it risks missing
a deeper opportunity: to treat AI not merely as a tool to regulate but as a subject of critical
learning.
Generative artificial intelligence (AI) tools, such as ChatGPT, Claude, and Google’s Gemini,
have rapidly become commonplace in daily life and are now permeating educational settings.
Many students already incorporate these systems into their study routines. However, they often
use them merely to expedite their assignments, accepting AI-generated answers at face value
without critical examination. This uncritical reliance on AI can lead to superficial learning
outcomes and a lack of deep understanding of the subject matter.
As a result, educators face a critical challenge: how to foster thoughtful and meaningful student
engagement with these tools. Rather than banning these systems or ignoring their influence, a
more constructive strategy is to integrate AI into the learning process as an object of critical

15
inquiry. Students should be encouraged to actively test and evaluate AI-generated responses. For
instance, they might cross-check the AI’s outputs against reliable sources or their own knowledge
and reflect on the AI’s performance (considering its accuracy, biases, and limitations). Through
this process, generative AI tools can be transformed from mere conveniences into catalysts for
deeper learning. Such an approach has the potential to sharpen students’ critical thinking skills,
deepen their comprehension of academic content, and cultivate a greater understanding of the
technologies that are increasingly shaping their education.
2. STUDY AIM AND SCOPE
This study investigates the educational potential of contemporary chatbots, with a focus on their
application in both direct student engagement and as support tools for instructors, particularly in
tasks such as generating assignments, designing assessments, and evaluating student submissions.
Given the rapid advancement in language models and digital technologies, it is plausible to
hypothesize that future developments may enable general-purpose chatbots to assume
responsibilities across the entire educational cycle. Such a vision envisions an advanced
pedagogical AI system capable of:
 Designing curriculum frameworks and course structures.
 Delivering instructional content interactively.
 Facilitating the development of problem-solving abilities and addressing conceptual
queries.
 Monitoring and assessing student progress through formative and summative evaluations.
 Generating final grades by integration with centralized academic records systems.
This leads to the conceptualization of a theoretical “ideal chatbot”—a fully autonomous digital
educator capable of replicating the core functions of human teaching professionals across all
stages of instruction.
However, despite significant progress in generative AI, particularly in natural language
understanding and the creation of textual, visual, and interactive educational resources, current
chatbot technologies fall considerably short of this ideal. Notable limitations include:
 Inability to perform long-term planning aligned with academic schedules.
 Absence of multi-user or group-based communication frameworks within the chatbot
interface.
 Limited capacity for contextual pedagogical adaptation, including real-time responsiveness
to individual learner needs.
 Weaknesses in interpreting ambiguous tasks and a lack of mechanisms for accountable
error recognition.
 Deficiencies in ethical and interpersonal decision-making—key dimensions of effective
teaching practice.
This study aims to explore the current capabilities and limitations of general-purpose chatbots
through practical experimentation, providing a systematic evaluation of their suitability for
educational use. The analysis identifies both the technological constraints and the pedagogical
risks associated with their deployment and proposes recommendations for their ethical and
meaningful integration into contemporary learning environments.
Preliminary findings suggest that, at present, it is premature to recommend chatbots for
autonomous use by students, even as personal learning assistants. Despite their sophisticated

16
linguistic capabilities, these systems remain prone to inaccuracies, including the phenomenon of
“AI hallucinations,” whereby outputs may appear plausible yet are factually incorrect, often in
critical or high-stakes contexts.
The study also briefly examined the potential role of chatbots in automating instructional design
tasks, such as generating quizzes, learning materials, and examination content. These outputs
could, in principle, be exported into formats compatible with widely used learning management
systems such as Blackboard or Brightspace [10]. Nevertheless, this paper primarily concentrates
on the implications of student–chatbot interaction, rather than instructor-focused applications.
3. TRANSFORMING LIMITATIONS INTO LEARNING: CRITICAL EVALUATION
OF GENERATIVE AI OUTPUTS IN INTERDISCIPLINARY LAB WORK
Considering the persistent error rates exhibited by large language models—rates that remain
considerable for reliable use in formal educational settings—this study proposes reframing such
limitations as pedagogical opportunities. Specifically, the development of laboratory-based
activities is suggested, wherein students critically examine chatbot-generated outputs, including
written responses, explanations, and problem solutions. By identifying inaccuracies and assigning
evaluative scores to these outputs, students not only refine their critical thinking skills but also
achieve a more nuanced and thorough grasp of the subject content.
Expanding on this approach, the paper introduces the concept of a novel interdisciplinary
instructional format: the Laboratory on Critical Evaluation of GenAI Outputs. This model shifts
the emphasis from the use of AI as a passive instructional tool to its role in cultivating essential
academic competencies. These include scientific rigor, analytical reasoning, and the capacity to
navigate and assess potentially unreliable or misleading information—skills increasingly vital in
today’s AI-augmented learning environments.
4. PEDAGOGICAL DESIGN OF A LABORATORY ON EVALUATING
GENERATIVE AI IN SUBJECT-SPECIFIC CONTEXTS
A defining characteristic of the proposed laboratory exercise is its implementation at the
culmination of a course, when students have already acquired foundational knowledge and core
competencies in the subject area. The primary aim is to assess the performance of various
generative AI chatbots in relation to the specific content of the course, encouraging students to
adopt a critical stance toward the quality and reliability of AI-generated information.
This laboratory task centers on active student engagement in the construction and analysis of
prompts directed at general-purpose language models. Students are required to independently
formulate a set of questions that span essential aspects of the course content and submit these to
multiple chatbot systems. The dual purpose of this exercise is to assess students’ mastery of the
material and to cultivate their ability to evaluate the interpretative capabilities and limitations of
GenAI tools within a disciplinary framework.
Crucially, students are guided to develop questions that differentiate chatbot responses by
degrees of correctness, ranging from wholly accurate to partially accurate to clearly erroneous.
To achieve this, students must:
 Review and internalize previously covered course material.
 Design questions of varying complexity and specificity.
 Develop evaluative criteria to assess the quality of chatbot-generated responses.

17
The task must be carefully calibrated: questions should avoid being overly simplistic (which may
result in uniformly correct responses from all systems) while also steering clear of highly
ambiguous or excessively specialized formulations that might falsely suggest chatbot inadequacy.
Upon collecting responses, students analyze the output for accuracy, completeness, and
conceptual depth. Responses are categorized into three types:
1. Fully correct – responses that are factually accurate, logically coherent, and well-
articulated.
2. Partially correct – responses that exhibit minor omissions, reasoning gaps, or stylistic
flaws;
3. Incorrect – responses containing factual inaccuracies, contradictions, or misleading
interpretations.
To support a structured evaluation, students apply a multi-dimensional rubric, which may include
criteria such as:
● Terminological precision.
● Contextual appropriateness.
● Depth of explanation.
● Logical coherence.
Through this analytical process, students develop critical thinking skills, reinforce their
understanding of course content, and gain practical experience interacting with GenAI tools.
Moreover, the activity fosters an interdisciplinary perspective on the role of AI in education by
highlighting both the capabilities and limitations of language models in content-specific
applications.
The exercise concludes with a comparative ranking of chatbot performance based on the
aggregated scores assigned by students, offering an evidence-based assessment of their relative
competence within the course domain.
5. EXPANDING INTERDISCIPLINARY GENAI LABORATORY WORK: A PILOT
IMPLEMENTATION IN ASTRONOMY EDUCATION
We propose the development of interdisciplinary laboratory activities that leverage generative
artificial intelligence (GenAI) across a range of academic disciplines. As an initial
implementation, this pedagogical model has been introduced within a general astronomy course
designed for students from non-STEM backgrounds.
The structure of the laboratory work may take the form of a single, integrated assignment with
three interconnected components—text, image, and video—or as three discrete tasks, each
dedicated to one modality. This tripartite framework is intended to acquaint students with a
variety of generative tools, enabling them to interact not only with language-based models but
also with systems capable of producing visual and multimedia content.
Astronomy was selected for this pilot due to its inherently imaginative and open-ended nature,
which makes it particularly conducive to creative engagement with generative technologies.
Unlike more rigidly structured scientific disciplines, astronomy invites speculative exploration.
GenAI tools—ranging from textual to visual chatbots—allow students to construct original
space-related narratives, conceptualize hypothetical celestial phenomena, and generate fictional

18
cosmological scenarios without requiring strict adherence to current scientific paradigms. This
flexibility lowers the initial cognitive barriers associated with using GenAI, fostering greater
student enthusiasm and creative experimentation.
The approach supports multiple pedagogical objectives simultaneously: enhancing students’
digital and GenAI literacies, promoting interdisciplinary thinking, and cultivating competencies
in visual communication and critical information analysis. Furthermore, it helps students
reconceptualize AI not merely as an automation tool but as an active partner in both cognitive
and creative processes.
Looking ahead, this methodology will be extended to other disciplines, including physics,
chemistry, and biology, with appropriate modifications to reflect the epistemological and
instructional characteristics of each field. This expansion aligns with a broader vision of
transdisciplinary digital transformation in higher education, in which GenAI tools are integrated
thoughtfully into domain-specific learning processes.
6. IMPLEMENTATION OUTCOMES AND PEDAGOGICAL IMPACT OF AN AI-
INTEGRATED ASTRONOMY LABORATORY FOR NON-SCIENCE MAJORS
As part of an experimental educational initiative, a laboratory activity was designed and piloted
within the General Astronomy course for non-science students. The instructional materials
included a comprehensive methodological guide, detailed instructions for conducting the lab, and
integrated links to digital forms for documenting student results.
The laboratory activity was implemented across several student cohorts over multiple academic
semesters. Each session was allotted two academic hours. While this duration proved insufficient
for all students to complete all three components of the lab (text, image, and video tasks), a
notable number of participants demonstrated rapid comprehension and completed the full scope
of the assignment within the allocated time.
Particularly noteworthy were the outcomes from the second and third phases of the lab, which
involved generating visual representations of imagined astronomical objects and phenomena
using GenAI tools. These images and videos—interpreted as speculative extraterrestrial
scenarios—were showcased during class sessions and stimulated vibrant peer discussion. This
not only fostered heightened cognitive engagement but also promoted collaborative learning and
creativity.
Students who were unable to complete the full task during class were allowed to finalize their
projects independently. Final submissions, particularly images and videos, were voluntarily
uploaded to a dedicated Facebook group for the course
(https://www.facebook.com/groups/GeneralAstronomy). This online platform functioned as a
supplementary engagement space, enhancing student participation and extending the learning
community beyond the classroom. Many students exceeded the minimum requirements of the
assignment, displaying substantial initiative and academic curiosity.
Moreover, several individual projects produced through this laboratory activity were
subsequently adapted into poster presentations for the annual BMCC student research symposium
(BARS). Student participation in the symposium was marked by a high degree of motivation and
received positive feedback from both peers and faculty, further underscoring the educational
value of integrating generative AI into the learning experience.

19
7. AI-ENHANCED ASTRONOMY LAB MANUAL
Introduction: Artificial Intelligence in Astronomy
Artificial intelligence (AI) is playing an increasingly transformative role in the field of astronomy
and space exploration. Key applications include:
 Data Processing and Image Enhancement: Machine learning techniques are employed to
refine astronomical images from space telescopes (e.g., Hubble, James Webb), eliminating
noise and increasing clarity. Notably, AI was instrumental in enhancing the historic image of
the supermassive black hole in galaxy M87.
 Pattern Recognition and Classification: AI systems assist in categorizing galaxies, stars, and
exoplanets, improving both the efficiency and precision of classification tasks. These systems
can also detect patterns across vast astronomical datasets, enabling the discovery of
previously unknown phenomena.
 Spacecraft Autonomy and Hazard Prediction: AI contributes to spacecraft autonomy by
processing environmental data and predicting threats such as solar flares and debris fields.
These capabilities optimize mission planning and reduce dependence on ground control.
 Forecasting Cosmological Events: AI models trained on historical data can anticipate rare
events, such as gravitational wave emissions and supernovae, thereby guiding observational
efforts more effectively.
 Scientific Efficiency and Collaboration: By automating repetitive analytical processes, AI
enables scientists to focus on higher-level reasoning and hypothesis testing, accelerating
research and reducing operational costs.
As AI technologies continue to advance, their integration into astronomy is expected to drive
further scientific breakthroughs. In this laboratory module, students will explore how generative
AI (GenAI) can enhance understanding of key concepts in astronomy, cosmology, astrophysics,
astrobiology, and space exploration.
Defining Generative AI in an Educational Context
Generative AI refers to systems that autonomously produce original content—text, images, video,
audio, or code—based on user-defined prompts. These systems are powered by advanced
machine learning architectures such as large language models (e.g., GPT), generative adversarial
networks (GANs), diffusion models, and transformers. The use of GenAI tools in education
presents novel opportunities to simulate phenomena, process scientific data, and encourage
creative inquiry.
Lab Objectives and Educational Aims
The lab is designed to achieve the following:
 Understanding GenAI Capabilities: Enable students to experiment with and evaluate
various GenAI platforms in the context of astronomy.
 Developing Critical Analysis Skills: Encourage students to assess the reliability,
consistency, and limitations of AI-generated outputs across different systems.
 Enhancing Engagement: Foster active student participation through practical and creative
interaction with digital tools.
 Building Technological Proficiency: Cultivate essential skills in navigating AI tools
increasingly relevant to modern scientific research and academic workflows.

20
Expectations and Learning Approach
 Active Engagement: Students are expected to interact thoughtfully with GenAI systems,
completing each task attentively and reflectively.
 Critical Inquiry: Students should approach each activity analytically—interrogating AI
responses, identifying errors, and drawing reasoned conclusions.
 Collaborative Learning: Peer discussion is encouraged to build a shared understanding of
GenAI performance and facilitate collective problem-solving.
 Innovative Thinking: Students are invited to devise challenging, imaginative prompts that
test the limits of GenAI reasoning and enhance conceptual understanding.
Lab Components
1. Evaluating GenAI Text Platforms
Objective: Students will design astronomy-related questions, input them into multiple GenAI
chat platforms, and compare the accuracy and clarity of the responses.
Expected Learning Outcomes:
 Ability to craft precise, content-relevant questions.
 Comparative analysis of GenAI-generated answers across platforms.
 Increased familiarity with various AI tools for scientific inquiry.
Required Materials:
 Computer with Internet access
 Lab Report Template: Google Doc Link
 GenAI Platforms:
 ChatGPT
 Gemini
 Claude
 Microsoft Copilot (Bing)
Procedure:
 Devise ten astronomy-themed questions with known answers (e.g., from textbooks or
course materials).
 Input questions into each GenAI platform and evaluate the responses.
 Attempt to identify questions where different platforms yield divergent or incorrect
answers.
 Analyse performance by noting which platform produces the most accurate and helpful
responses.
Example Topics: Planetary order, celestial mechanics, notable space missions, moon phases,
stellar evolution, and observational astronomy.

21
2. Creating AI-Generated Astronomical Imagery
Objective: Students will explore GenAI image generators to visualize cosmic phenomena and
assess the realism and creativity of the generated outputs.
 Develop prompt engineering skills tailored to scientific visualization.
 Compare GenAI-generated images with reference astronomical imagery.
 Understand the creative liberties and scientific constraints of AI-based image generation.
Recommended Platforms:
 Night Cafe
 Bing Image Creator
 Google ImageFX
 Craiyon
Procedure:
 Create detailed prompts describing astronomical objects or events.
 Generate images using multiple platforms and compare outputs.
 Assess images based on scientific plausibility, visual fidelity, and imaginative appeal.
 Include images and descriptive prompts in the final lab report.
3. Visualizing Dynamic Astronomical Phenomena via GenAI Video Tools
Objective: Students will produce short educational videos illustrating dynamic space phenomena
(e.g., black hole mergers, eclipses, planetary motion) using AI video platforms.
 Ability to research and storyboard complex phenomena.
 Experience with GenAI video tools to communicate scientific content effectively.
 Improved science communication skills through multimedia storytelling.
Suggested Tools:
 InVideo
 Kapwing
 Synthesia
Procedure:
 Select a phenomenon and conduct brief background research.
 Write a concise script or storyboard outlining the visual sequence.
 Use GenAI video platforms to generate the visual content, adding narration or captions if
desired.
 Submit completed videos via Blackboard and provide a brief reflection on the creative
and technical process in the lab report.

22
Conclusion
This AI-enhanced astronomy lab empowers students to critically engage with generative
technologies, deepen their understanding of scientific concepts, and develop skills relevant to the
future of interdisciplinary education. By exploring AI’s strengths and limitations across textual,
visual, and audiovisual outputs, students are not only positioned as consumers of educational
content but also as informed and reflective creators who can leverage GenAI tools in their
academic and professional trajectories.
8. DISCUSSION
This project demonstrated that the integration of generative artificial intelligence (GenAI) tools
into laboratory-based instruction can significantly enhance student engagement, particularly
among learners from non-scientific academic backgrounds. The lab not only supported content
acquisition in astronomy but also encouraged the development of critical inquiry, digital literacy,
and confidence in interacting with emerging technologies.
8.1. Strengths and Pedagogical Benefits
Student engagement throughout the lab activities was notably high. Learners responded
positively to the opportunity to interact with various GenAI platforms, using them to generate
text, visualizations, and video content. These multimodal tasks fostered creativity and enabled
students to explore complex scientific themes in a manner that was both accessible and
personally meaningful.
The comparative analysis of outputs from different GenAI tools further deepened students’
understanding of how these systems function, highlighting both their generative potential and
their limitations. The voluntary continuation of projects beyond the classroom, including sharing
outputs on digital platforms and presenting at student symposia, indicates a high level of intrinsic
motivation and meaningful engagement. Such active learning experiences are widely recognized
as more effective than passive instructional formats such as lectures or readings.
8.2. Challenges and Considerations
Despite these successes, the implementation of GenAI in education raises several challenges. One
of the most pressing concerns is the potential for students to accept AI-generated responses
uncritically. GenAI systems, such as ChatGPT, may produce responses that appear coherent and
plausible but are factually inaccurate or misleading. Without sufficient guidance, students may
internalize incorrect information, compromising their understanding of scientific concepts.
It is therefore essential that educators provide explicit instruction in critical evaluation skills and
encourage students to verify AI-generated content against authoritative sources. Moreover,
students must be taught to understand that GenAI systems do not possess knowledge in a human
sense but operate probabilistically, generating responses based on statistical patterns in language
data rather than verified facts.
8.3. Toward More Informed and Reflective Use of GenAI
The lab experience contributed to a pedagogical shift in students’ approach to AI tools—from
reliance on GenAI for quick answers to a more reflective and analytical engagement. Students

23
learned to differentiate between high-quality, incomplete, and erroneous responses and developed
skills in crafting effective prompts and evaluating the educational value of AI outputs.
Overall, the project underscores the potential of GenAI to serve as a constructive educational
resource when integrated thoughtfully and supported by appropriate scaffolding. With educator
guidance, GenAI can move beyond being a passive content generator to becoming a catalyst for
deeper thinking, creativity, and digital competence in 21st-century learners.
8.4. Rethinking the Limitations of the Study Considering AI-Student Interaction
If we judge this study by the usual rules of academic research, it might seem weak. There was no
control group, no standardized test to measure learning, and no way to repeat the same
experiment. But these aren’t just mistakes — they reflect a new and unusual way for students to
interact with AI.
Generative AI tools like ChatGPT don’t always give the same answer to the same question. They
answer differently depending on how and when the question is asked. Because of this, using
rigorous research methods doesn’t always make sense. Trying to control everything would
deprive students of the opportunity to explore and learn naturally.
Instead, this study allowed students to freely try things out and learn by doing. And it revealed
some important things: students learned to ask better questions, check the AI’s answers, and think
more deeply when the AI made mistakes. They even continued to work with the AI after the
session ended, which shows real interest and curiosity.
To better understand this type of learning in the future, we need new ways to study it. We need to
look at how students use AI over time, what questions they ask, how they respond to incorrect
answers, and how their thinking changes. Instead of focusing solely on test scores, we need to
look at how students develop their ability to use and question AI effectively and intelligently.
In short, while this study may not follow all the usual rules of educational research, it opens the
door to a better understanding of how students can learn and teachers can teach in today’s world,
using often-obscure AI as a tool and as a learning partner.
9. DIFFERENTIATING THE EDUCATIONAL UTILITY OF TEXT, IMAGE, AND
VIDEO-BASED GENERATIVE AI TOOLS
Our evaluation of currently available generative artificial intelligence (GenAI) platforms
indicates notable differences in their educational applicability, depending on the modality of
content generation—text, image, or video.
Text-based GenAI tools, such as ChatGPT and Claude, have demonstrated considerable
pedagogical value. These systems excel in delivering detailed explanations, structured reasoning,
and context-rich narratives that support the comprehension of both foundational and advanced
theoretical concepts. When tested against questions of typical undergraduate textbook difficulty,
text-based GenAI systems produced accurate responses in over 90% of cases. However,
occasional inaccuracies—often presented in a confident and linguistically plausible manner—
highlight the need for critical oversight. The conversational interface of these tools fosters student
engagement by enabling iterative questioning and immediate clarification, while the ability to
compare outputs from different platforms supports the cultivation of critical evaluation skills.

24
Visual GenAI tools, including platforms such as NightCafe and Craiyon, offer the potential to
enhance learning through visual representations, which are often more memorable and accessible
than textual explanations. Nonetheless, at their current stage of development, these platforms
exhibit significant limitations in scientific accuracy. Empirical testing indicates that graphical
GenAI systems fail to produce reliable illustrations or diagrams in response to prompts involving
college-level scientific or technical content, with error rates approaching 100% in such contexts.
Video-based GenAI systems, such as Synthesia and InVideo, are similarly promising for
depicting dynamic processes and enhancing engagement through audiovisual storytelling.
However, their outputs currently lack the scientific precision necessary for accurate
representations in domains requiring factual rigor, such as physics, chemistry, or biology.
Considering these findings, we propose that text-based GenAI tools are well-suited for immediate
integration into a broader range of college-level science curricula, including physics, chemistry,
and biology, where conceptual clarity and verbal explanation are central to student
understanding.
In contrast, graphical and video-based GenAI tools are currently best suited to creative,
exploratory educational contexts, such as the speculative modelling of extraterrestrial
environments in astronomy. In these settings, where imaginative engagement is prioritized over
factual accuracy, visual GenAI can effectively enhance student interest, promote creative
thinking, and support interdisciplinary learning without the risk of disseminating misinformation.
Nevertheless, the broader application of image and video GenAI in science education,
particularly where accuracy and detail are paramount, remains constrained. Progress in this area
will depend on future advancements in GenAI capabilities and/or the development of targeted
pedagogical frameworks that can safely integrate these tools without compromising scientific
integrity.
REFERENCES
[1] Extance, A. (2023, November). ChatGPT has entered the classroom: how LLMs could transform
education. Nature, 623(7987), 474–477.
[2] Cukurova, M. (2024). The promise and challenges of generative AI in education. Behaviour &
Information Technology, 43(1).
[3] Isotani, S. (2024, January). Harnessing AI's Powers For All: Exploring how behavioral science and
technology can make a positive impact on education. Harvard Graduate School of Education.
[4] Lee, S.-A., Welch, J., Wallace, R. J., Cross, D., & Loffi, J. M. (2024). ChatGPT in the Classroom:
A Practical Guide for Educators. Journal of Aviation/Aerospace Education & Research, 33(3), 1–
31.
[5] Koch, L. (2023). ChatGPT in the Classroom: Uses, Limitations, and Student and Teacher
Experiences. The Journal of Adventist Education, 85(3), 4–10.
[6] Farrokhnia, M., Banihashem, S. K., Noroozi, O., & Wals, A. (2023). A SWOT analysis of
ChatGPT: Implications for educational practice and research. Innovations in Education and
Teaching International, 61(3), 460–474.
[7] Chiu, T. K. F., Xia, Q., Weng, X., & Ouyang, F. (2024). A scoping review on how generative
artificial intelligence transforms assessment in higher education. International Journal of
Educational Technology in Higher Education, 21(1).
[8] Kohnke, L., & Moorhouse, B. L. (2025). Enhancing the emotional aspects of language education
through generative artificial intelligence (GenAI): A qualitative investigation. Computers in Human
Behavior, 167, 108600.
[9] Oates, A., & Johnson, D. (2025). ChatGPT in the Classroom: Evaluating its Role in Fostering
Critical Evaluation Skills. International Journal of Artificial Intelligence in Education.
[10] Znamenskiy, V. (2023, May 12). Trying to get the ChatGPT AI to generate educational quizzes in a
specific format designed for input to Blackboard. 2nd Annual BMCC Assessment Day Colloquium.

25
AUTHOR
Dr. Vasiliy Znamenskiy, who holds a Ph.D. in Physics and Mathematics, teaches physics, astronomy,
computer methods in science, and computer-aided analysis for engineering at CUNY (BMCC and CITY
TECH). His interests include molecular dynamics simulations, quantum chemistry, hydrogen bonding,
neuromorphic systems, and AI-enhanced science education. He integrates computational techniques and AI
into both student learning and scientific research.

Integrating Universal Generative AI Platforms in Educational Labs to Foster Critical Thinking and Digital Literacy

More Related Content

Similar to Integrating Universal Generative AI Platforms in Educational Labs to Foster Critical Thinking and Digital Literacy (20)

Recently uploaded (20)

Integrating Universal Generative AI Platforms in Educational Labs to Foster Critical Thinking and Digital Literacy