Promoting Critical Thinking in Biological Sciences in the Era of Artificial Intelligence: The Role of Higher Education

Abstract

The integration of artificial intelligence (AI) into the biological sciences marks a transformative era, reshaping research methodologies, data analysis, and hypothesis generation. This technological advancement accelerates discoveries and enhances our understanding of complex biological systems. As AI increasingly influences decision-making processes, the necessity for students and scientists to critically assess AI-generated outputs becomes paramount. The current narrative review explores the evolving role of critical thinking in biological sciences amidst the rise of AI, emphasizing the importance of skepticism, contextual understanding, and ethical considerations. It argues that while AI provides powerful tools for data interpretation and pattern recognition, human oversight and critical analysis remain indispensable to validate findings and prevent biases inherent in automated systems. Higher education institutions play a crucial role in fostering a culture of critical thinking, equipping biological scientists to effectively harness AI technologies while ensuring the integrity of their research and upholding scientific and ethical standards. Furthermore, AI tools, including chatbots, could be strategically employed in active learning methodologies, such as problem-based learning, flipped classrooms, and online learning. These methodologies enhance the ability of students to effectively utilize AI technologies while ensuring the rigor of scientific research. In conclusion, the current review underscores the benefits, challenges, and educational implications of AI integration, offering actionable insights for educators and learners seeking to adapt effectively to this rapidly evolving technological landscape.

1. Introduction

In an era where artificial intelligence (AI) is revolutionizing the ever-expanding fields of biological sciences, the need for critical thinking (CT) has never been more urgent [1]. Scientists in this field frequently encounter complex, data-intensive subjects that require the application of both theoretical and practical knowledge, and often rely on AI to analyze and interpret [2]. Given how rapidly AI technologies are transforming scientific research, it is logical for higher education institutions to embed AI into their teaching frameworks, enabling the next generation of scientists to gain relevant expertise. AI offers promising enhancements, such as personalized learning experiences and advanced data analysis tools, but it also presents unique challenges [3]. Educators must navigate the delicate balance between leveraging AI to facilitate learning and ensuring that students remain at the forefront of the critical evaluation process [4]. So far, we have made strides in promoting CT through educational methods, including problem-based learning and laboratory-based learning [5]. With AI entering the arena, we need to start re-defining CT skills in this new context. The ability of these educational tools and methodologies to develop CT skills will largely depend on their capacity to foster an environment that encourages questioning and openly addresses cognitive biases.

Besides biological sciences, the development of CT skills remains a universal goal among educators, becoming even more critical considering recent crises, such as climate change and the COVID-19 pandemic [6]. These events underscore the urgent need for individuals to assess information critically and make well-informed decisions. Additionally, in a post-truth era characterized by compromised fact-gathering processes, the necessity for CT across all sectors of society becomes even more pronounced [7]. CT is indispensable, transcending academic confines of science education or philosophy, as it equips individuals to navigate the complexities of a modern, information-rich world [8].

The cultivation of CT within higher education requires innovative pedagogical strategies that go beyond transmitting knowledge [9]. These strategies should challenge students to question assumptions and refine their analytical abilities. Exploring diverse teaching methods is essential to meet the evolving needs of students and to create more engaging and interactive educational environments. In this context, higher education serves as a crucial foundation, setting the stage for future academic and professional success [5].

AI-powered tools can enhance traditional classroom strategies by offering personalized learning materials, automating routine assessments, and enabling immediate feedback on student work. Using the right prompts, interaction with AI-driven applications encourages learners to critically appraise data outputs, prompting them to question data accuracy, potential biases, and the relevance of conclusions to real-world contexts. These rely on the educator effectively using the tools with guided reflection and collaborative projects. In this context, students must interpret or refine intelligent models, and AI can help refine problem-solving abilities and deepen students’ appreciation for evidence-based reasoning. This approach equips students with the confidence to navigate complex information ecosystems and fosters an enduring commitment to analytical thinking [10,11].

In this review, we explore the pivotal role of higher education in fostering CT in the biological sciences during the era of AI in several ways:

• Firstly, we examine how to adapt curricula effectively to integrate AI technologies, thus preparing students to fully engage with both the functionalities and limitations of these systems.
• Secondly, we outline a blend of pedagogical approaches, providing a framework for developing a balanced educational setting. Here, the strengths of AI are leveraged strategically, with an equal or greater emphasis on ensuring that scientific rigor and reflective practices remain paramount.
• Thirdly, we underline the importance of preparing future biological scientists to navigate an increasingly AI-driven research landscape in a way that prioritizes both ethical considerations and evidence-based inquiry. Through discussions of algorithmic biases, data vulnerabilities, and accountability, we address how students can develop a heightened awareness of the societal and scientific implications of automated decision-making processes.
• Fourthly, as part of creating a well-rounded learning experience, we highlight the value of advanced tools, such as adaptive learning platforms and automated assessments, to reinforce student engagement and promote iterative learning.
• Lastly, we pave the way for designing learning experiences that cultivate shared investigation skills, bolster cross-disciplinary cooperation, and equip graduates to respond to emerging challenges in big-data biology. In doing so, we position institutions to form the next generation of researchers who can apply AI solutions responsibly, drive innovation, and maintain a skeptical, yet constructive stance in the face of rapidly evolving scientific technologies.

The current paper is a narrative literature review. To ensure coverage of relevant and current scholarship, we conducted a targeted literature search using PubMed, Scopus, and ScienceDirect databases up to February 2025. Search terms included combinations of “critical thinking”, “artificial intelligence”, “machine learning”, “higher education”, and “biological sciences”, using Boolean operators AND and OR to refine results. Additional keywords, such as “virtual labs”, “tutoring systems”, and “tutoring tools”, were used in combination with “biological sciences” and “higher education” to identify relevant tools and pedagogical strategies. Although no formal inclusion or exclusion criteria were applied, we prioritized peer-reviewed articles, reviews, and commentaries that directly addressed AI’s role in promoting CT within educational settings—particularly in the context of the biological sciences. In keeping with the flexible nature of a narrative review, after our initial search, we reviewed each article’s abstracts and relevant full sections to identify content on AI-enhanced teaching, learning activities, or references to CT in biological sciences. Points related to educational design, pedagogical practices, and ethical concerns were noted, taking particular care to capture authors’ perspectives on how AI might shape or challenge CT. These informed our discussion and guided our recommendations for integrating AI into biology curricula.

2. Developing Critical Thinking in Biological Sciences and the Role of Higher Education

2.1. Defining Critical Thinking

Though commonly attributed to Socrates, and the term ‘critical thinking’ popularly linked to John Dewey’s writings in 1910 [12], recent scholarship offers a broader historical perspective. It has been suggested that critical reflection predates Socrates, tracing back to the Presocratic philosophers who initially pioneered systematic questioning and skepticism [9]. Moreover, the use of ‘critical thinking’ in scholarly texts predates Dewey, suggesting a more diverse and rich origin of the concept than traditionally recognized.

CT has been considered a cornerstone of scientific study and higher learning for many decades. While Dewey’s earlier writing addressed reflective thinking as central to education, later frameworks built upon more clearly worded and operationalized definitions of CT applied to educational and professional environments. Perhaps strongest of all is the definition offered by Peter Facione, whose Delphi Report (available at: https://eric.ed.gov/?id=ED315423; accessed on 2 March 2025) directed much of contemporary scholarship on CT. Facione defines CT as “purposeful, self-regulatory judgment” that involves interpretation, analysis, evaluation, inference, and explanation. Importantly, he also appeals to the dispositional components of CT—open-mindedness, inquisitiveness, and commitment to truth-seeking—which are especially crucial in scientific disciplines where ambiguity, complexity, and shifting evidence reign (discussed further below). Similarly, Robert Ennis has provided an extensively used definition of CT as “reasonable reflective thinking focused on deciding what to believe or do” [13]. Ennis emphasizes the teachability of CT as a primary skill and advocates for embedding it across disciplines through direct instruction, practice, and feedback. His taxonomy of CT skills, such as clarifying problems, evaluating source credibility, and recognizing inconsistencies, provides a practical structure that is apt for the aims of biology education and the moral use of AI technologies. Embedding these traditional models allows more formalized understanding of CT, particularly applicable within AI-mediated learning. As students increasingly use AI-mediated learning devices that generate and interpret information, they must be equipped not only with cognitive abilities but also with reflective tendencies to question, contextualize, and ethically evaluate AI-generated information.

CT transcends mere academic skill—it embodies a lifestyle, cultivated through experiential learning rather than direct instruction. It involves an active engagement with information, where learners critically assess and integrate knowledge through active learning processes [14]. Historically, CT involves the application of good reasoning to make informed decisions, a foundational aspect of the scientific method. This method relies on both inductive and deductive reasoning, as scientists formulate hypotheses, analyze evidence, and develop comprehensive theories about the natural world.

CT is defined as an encompassing analysis, evaluation, and the formation of inferences, underpinned by a flexible mindset [15]. A scientifically literate individual should not only possess knowledge of specific scientific content but also be adept at employing CT and associated reasoning skills to evaluate the validity of daily scientific claims. Furthermore, the Delphi Report on CT (https://eric.ed.gov/?id=ED315423; accessed on 1 March 2025) outlines essential cognitive skills integral to this process:

i.

Analysis: dissecting concepts or ideas into smaller components to understand their structure and relationships.

ii.

Inference: drawing conclusions by reconciling known facts with unknown elements.

iii.

Evaluation: assessing evidence to form reasoned judgments within a specific context.

Additional skills pertinent to scientific inquiry include interpretation, explanation, and self-regulation. According to the statement of Expert of Consensus for Purposes of Educational Assessment and Instruction (https://eric.ed.gov/?id=ED315423; accessed on 2 March 2025), CT also involves behavioral tendencies or dispositions, such as pursuing truth, open-mindedness, analytical thinking, systematic organization, and inquisitiveness. These dispositions align closely with behaviors esteemed in scientific practice, highlighting the direct benefits of emphasizing CT in science education.

Current curricular and educational policy documents at all levels of education widely acknowledge CT. Industry and the labor market are also giving it increasing recognition [16]. The advent of new technologies and automation highlights the importance of abilities that AI or algorithms cannot perform [17]. CT thus emerges as one of the key soft skills and 21st-century competencies, encompassing creativity, flexibility, fairness, empathy, communication, cooperation, social awareness, and context-based adaptation. Without these skills, technology could become detrimental to society [18]. However, new technological developments could be leveraged to advance CT more comprehensively and enhance the efforts of education systems, as their current impact remains insufficient [19].

By fostering these cognitive and behavioral aspects, educators can enhance students’ ability to engage meaningfully in scientific inquiry, preparing them for complex problem-solving and informed decision-making in their academic and professional pursuits

2.2. Current Educational Practices That Promote Critical Thinking

CT is a primary skill in undergraduate learning—the capacity to use data and evidence in making educated judgments about what to believe and what to do. It is not only a central learning goal across courses but also a requirement in an information-rich world where information is readily available. Besides, employers value CT to a high extent, regarding it as the tool for career success, which entails employees’ decisions based on facts, thoughts in different dimensions, and accepting individual limitations [20].

Innovative pedagogies have proven to possess considerable power in improving learning outcomes, as well as enhancing students’ CT capacity. Some of these effective practices are summarized in

Table 1. Pedagogical approaches to promote critical thinking in education.

Teaching Method	Description	Ref.
Inquiry-Based Learning	Encourages students to actively engage with course material through questioning, finding answers, and problem-solving and investigative activities, enabling clearer understanding.	[21]
Problem-Based and Project-Based Learning	Involves students in complex, real-world projects or problems over an extended period, enhancing problem-solving skills, analytical abilities, and promoting teamwork and innovative thinking.	[22]
Gamification	Incorporates game elements into learning to improve engagement, motivation, and critical thinking, making the learning process more enjoyable and deeply engaging.	[23]
Cooperative Learning	Students work in small teams toward educational goals, enhancing critical thinking by exposing them to different viewpoints and collaborative problem-solving.	[24]
Experiential Learning	Links theory to practice by allowing students to apply knowledge in real-world contexts, such as internships and fieldwork, thus enhancing understanding and practical application of theories.	[25]
Community-Engaged Learning	Integrates community service projects with coursework to enhance problem-solving skills and demonstrate the real-life relevance of academic studies.	[26]
Design-Based Research	Supports the development of teaching strategies that foster critical thinking through iterative cycles of design, enactment, analysis, and redesign.	[27]
Flipped Classrooms	Allows students to learn foundational knowledge at home with pre-recorded lectures and apply this knowledge in class through interactive and collaborative activities for deeper learning.	[28]

.

There is evidence in the literature of educational research that supports the effectiveness of these dynamic models to greatly increase CT capacity [29]. For example, problem-based learning (PBL) positions students in actual problems, more fully allowing them to apply higher-order thinking skills [30]. Similarly, flipped classrooms allow for more in-depth coverage of the content, challenging students to move beyond mere memorization into analytical and creative forms of thought. With the application of these varied approaches, instructors can create a rich, engaging, and effective learning setting that not only transfers knowledge but also fosters the CT skills necessary for students’ success in school and the workplace.

2.3. Biological Sciences and the Need for Critical Thinking: From Hypothesis Formation to Scientific Innovation

The field of biological sciences is the branch of science focused on the study of living organisms, their structures, functions, growth, evolution, and interactions with the environment. This expansive domain encompasses topics ranging from the molecular mechanisms within cells to ecological networks spanning entire ecosystems. To name a few fields, biological sciences include molecular biology, cell biology, genetics and genomics, biochemistry, ecology, evolutionary biology, physiology, microbiology, immunology, and neuroscience. Each of these fields is vast and complex. To add to the complexity, these fields frequently collaborate while drawing insights from interdisciplinary methods in chemistry, physics, computer science, and mathematics to address scientific questions.

This collaborative approach aids in generating comprehensive solutions for complex biological challenges and drives science forward at an incredible pace. At the same time, it requires higher-order thinking and analysis that may only be supported by CT. At the most basic level, sciences in general, and biological sciences in particular, follow the scientific method, which requires CT at every step. Figure 1 illustrates the cyclical nature of the scientific method, emphasizing the centrality of critical thinking at each stage of the process. From hypothesis formation to methodology selection, data analysis, and interpretation, scientific inquiry demands a continuous loop of evaluation, reflection, and refinement. This highlights how critical thinking is essential not only for data interpretation, but also from the initial questioning phase through to the decision of whether to accept, refine, or reject a hypothesis. By embedding critical evaluation throughout the cycle, Figure 1 demonstrates that critical thinking is a fundamental, non-linear component of scientific practice. As AI tools become increasingly integrated into each of these stages, it is crucial that biological sciences education explicitly cultivates these cognitive and evaluative skills.

The unique challenges and rapid developments in this field necessitate a robust capacity for critical evaluation and creative problem-solving among professionals and students alike. Below, we examine each stage of the process and discuss the need for CT:

i.

Hypothesis development and testing. CT enables scientists to systematically develop and scrutinize hypotheses, a fundamental aspect of the scientific method [31]. This is especially crucial in fields like behavioral ecology, where researchers formulate and test hypotheses regarding animal behavior, such as the “dear enemy hypothesis”, which explores how territorial animals react less aggressively to known neighbors than to strangers [32].

ii.

Data analysis and interpretation. In biological research, CT is indispensable for analyzing complex data accurately. It ensures that scientists draw logical conclusions from their findings, which is essential for advancing knowledge and making informed decisions [33].

iii.

Navigating complex information. The exponential growth in biological knowledge and data requires the ability to critically assess vast amounts of information and distinguish between valid scientific evidence and misleading or incorrect claims. This skill is vital for researchers to navigate through the complexity and integrate new information effectively [34].

iv.

Scientific inquiry and innovation. CT is the engine driving scientific inquiry and innovation. It encourages scientists to challenge existing assumptions and explore alternative explanations, leading to new discoveries and advancements in the field.

v.

Evaluating biological claims. The ability to evaluate and critique biological claims critically is increasingly essential. For instance, in educational settings, students can benefit from exercises that involve assessing the validity of scientific claims, enhancing their understanding and CT skills. Such pedagogical approaches not only reinforce the scientific content but also prepare students to think critically in real-world scenarios [35].

vi.

Impact of inquiry-based learning. Inquiry-based learning strategies in laboratory courses have shown to significantly improve CT skills. These strategies involve students in active learning through research and problem-solving, which better prepares them for professional challenges. Implementing these approaches in education can transform passive learning environments into dynamic platforms where CT thrives [36].

It thus becomes apparent that at each level, CT is required, as it involves evaluating the credibility of sources, identifying gaps in existing knowledge, and formulating logical conclusions based on evidence. In the biological sciences, this entails taking time to scrutinize experimental design, use appropriate statistical tests, and remain open to revising hypotheses if new findings emerge. By practicing these habits, students and researchers can mitigate erroneous interpretations of data and uphold the integrity of the scientific process. Consider two scenarios. Imagine a researcher analyzing a large-scale gene expression dataset and they notice unexplained outliers. Applying CT, they would investigate the sample preparation methods, sequencing quality metrics, or any discrepancies in data collection protocols before drawing conclusions about changes in gene regulation. In another case, a researcher in an ecological field study may discover an unusual shift in a species’ population dynamics. Rather than attributing the phenomenon to a single factor, CT prompts them to examine multiple influences, such as habitat loss, climate variability, or possible measurement errors. This approach helps generate a more nuanced and accurate explanation of the observed pattern.

2.4. The Challenges of Development of Critical Thinking in the Artificial Intelligence Era

Despite its importance, formal guidance for cultivating CT through emerging AI technologies can be insufficient, prompting the need to build on proven, hands-on techniques. Methods such as inquiry-based learning and problem-based or project-based activities leverage students’ natural curiosity while reinforcing analytical oversight. Gamification and cooperative learning add interactive dimensions that encourage engagement, while experiential and community-engaged learning provide real-world application to support deeper reflection on both the process and outcomes. Design-based research and flipped classroom structures further boost critical examination by shifting core content mastery to individual study and freeing class time for collaborative debate. As educational institutions integrate AI-driven assessments and adaptive tools, these pedagogies should evolve to ensure students remain active participants in the learning process, frequently evaluating algorithmic outputs and their ethical implications, rather than simply accepting them at face value.

In practice, this might include group-based data evaluations that integrate hypothesis-driven inquiry and AI-powered tools, accompanied by transparent rubrics that highlight potential biases or missing perspectives. By embedding these reflective elements within computer-assisted modules, educators can uphold high standards of critical evaluation, while still embracing the efficiency and scalability AI brings to modern learning environments. In the next section, we discuss the contribution of AI in education at present.

3. Artificial Intelligence in Education

AI in education (AIEd) is not a new concept. In fact, it has been around for over 30 years. Given the involvement of AI and education in knowledge-intensive cognitive development, their collaboration is a natural step toward shared objectives. As such, it is not surprising that numerous applications have emerged to enhance learning [37].

With increasing interest in AI and education, the Education 4.0 Alliance sought to understand the current state and future promises of the technology for education. The latest report, titled “Shaping the Future of Learning: The Role of AI in Education 4.0”, shows four key promises that have emerged for AI to enable education 4.0: (i) supporting teachers’ role: augmentation and automation, (ii) refining assessment and decision-making in education, (iii) supporting AI and digital literacy, and (iv) personalizing learning content and experience (https://initiatives.weforum.org/reskilling-revolution/education-4-0; accessed on 28 February 2025).

Globally, the increasing demand for national education reforms and the need to alleviate the burdens on educators have spurred the adoption of AI technologies. This shift has led to the widespread introduction of educational AI tools that support various aspects of teaching and learning [38]. In the 21st century, as we delve deeper into the Fourth Industrial Revolution (4IR or Industry 4.0), the application of AI has introduced advanced intelligent tutoring systems (ITSs). These systems are designed to mimic and effectively replicate the functions of human tutors. In their recent systematic review, Wang et al. [37] identified the main categories and subcategories where AI is used in education. Specifically, the main four categories they noted were adaptive learning (including personalized tutoring), intelligent assessment and management, profiling and prediction, and finally, emerging technologies.

Moreover, one of the hallmark achievements of Industry 4.0 is the development of machines that emulate the cognitive and perceptual abilities of biological systems—capabilities like object recognition and decision-making, which are fundamental to human intelligence [39]. These advancements have wide-reaching implications, extending into fields such as medical research, agriculture, and bio-based industries, all of which play pivotal roles in promoting sustainable lifestyles.

The following section discusses the technologies supporting CT in biological sciences.

3.1. Adaptive Learning Through the Use of Intelligent Tutoring Systems

An ITS typically represents a computer-based learning platform that employs AI algorithms to guide students toward deeper understanding. Often, these algorithms adjust content delivery, pose questions, and offer feedback by considering each learner’s individual mastery level and learning style [40]. ITS models can initiate conversational exchanges through natural language processing between an animated avatar and the learner. Currently, the technology has been progressing to systems that imitate human speech and facial gestures. Fundamentally, the ITS aims to mirror the supportive environment of an in-person tutor and has been shown to be effective in STEM subjects [41]. AutoTutor (https://autotutor.org/), initially used by physics students, is an ITS designed to engage students in conversational dialogues that prompt deeper understanding of academic topics. It uses natural language processing to interact with learners through animated avatars, providing targeted hints, corrective feedback, and adaptive questioning. This approach can help students refine their reasoning skills over time by closely emulating one-on-one human tutoring experiences [40]. While newer tutoring systems have also emerged, many educational technology researchers incorporate AutoTutor’s approach to conversational interaction when experimenting with adaptive systems and automated feedback.

In the current landscape of ITS, researchers continue to develop newer solutions that integrate enhanced machine learning methods, advanced analytics, and conversational agents. These agents are automated software programs capable of understanding and responding to student inquiries using natural language. They can simulate dialogues, mimic human gestures through animated avatars, and even offer real-time hints based on learner profiles. For example, iTalk2Learn incorporates speech recognition in mathematics education, allowing the tutoring agent to adapt feedback strategies in accordance with students’ spoken responses [42]. Additionally, the Generalized Intelligent Framework for Tutoring (GIFT) provides a flexible architecture for building agent-based ITSs customizable to multiple subject areas. These emerging platforms demonstrate how agents, paired with deeper modeling of student behavior, can provide more nuanced support. By engaging learners in interactive exchanges, they foster a setting where students are guided to reflect on mistakes, reexamine assumptions, and remain active participants in the learning process.

Newer ITSs include platforms and frameworks that integrate advanced machine learning and adaptivity to offer real-time personalization, as summarized in

Table 2. An overview of newer intelligent tutoring systems (ITSs).

ITS Platform	Description
DeepTutor https://deeptutor.knowhiz.us/, accessed on 1 May 2025	Utilizes deep natural language dialogues to assess students’ understanding and provide responses tailored to their learning needs.
iTalk2Learn https://www.italk2learn.com/, accessed on 1 May 2025	Focuses on mathematics education, adapting feedback based on speech recognition and the analysis of students’ problem-solving patterns.
Generalized Intelligent Framework for Tutoring (GIFT) https://gifttutoring.org/projects/gift/wiki/Overview, accessed on 1 May 2025.	Offers a flexible architecture that supports the creation of ITSs across various domains, incorporating advanced analytics and learner modeling.
ASSISTments https://new.assistments.org/, accessed on 1 May 2025.	Allows for iterative practice and immediate feedback with strong teacher oversight, enhancing the ability to efficiently monitor students’ claims and identify errors.
Deep Knowledge Tracing-based Systems [43]	Uses deep neural networks to predict future learner performance, enabling more accurate and tailored adaptive interventions.

.

Many university courses, including biology, bioinformatics, and other STEM courses, have benefited from the integration of ITSs. Systems such as EER-Tutor and MATHia provide learners with direct feedback on database modeling or mathematical problem-solving, enabling them to revisit and refine their approaches continuously [44]. ALEKS, which relies on knowledge space theory, has been used extensively in foundational science and mathematics courses to identify student knowledge gaps [44]. For those involved in biological research, tools like BioTutor (https://www.yeschat.ai/gpts-2OToOArHrZ--Bio-Tutor, accessed on 1 May 2025) offer adaptive questioning and scaffolding tailored to topics such as cellular biology and genetics. Larger-scale platforms, such as ASSISTments and DeepTutor, offer similarly robust feedback loops and performance tracking, making them valuable for students pursuing interdisciplinary biological research. VIPER, which has been used in certain computational biology contexts, supports learners by guiding them through complex modeling exercises, assisting with visualizing large molecular or genomic datasets. Similarly, COMET has been cited for its analytical coaching features, which prompt learners to articulate their reasoning steps while tackling intricate problems in fields like biostatistics or systems biology.

By employing these systems, educators can address individual learning gaps, provide personalized insights into complex workflows, and maintain data-driven approaches to student improvement. In particular, VIPER’s emphasis on problem-solving paths can be advantageous for students navigating advanced simulations or large data repositories, as it ensures that they understand each parameter’s impact on computational outcomes. COMET, on the other hand, blends analytical support with conceptual prompts, making it well suited to students who need to refine their thinking during steps such as formulating hypotheses or designing experimental protocols [45]. Both systems, when used responsibly in classrooms or research-based settings, reinforce valuable reflection and iterative learning processes. The adaptability of these ITS platforms holds promise for a range of learners, from those newly introduced to STEM fundamentals to more advanced students conducting cross-disciplinary analyses in computational biology.

3.2. The Role of Chatbots

Chatbots are interactive software applications powered by algorithms, often incorporating AI, designed to simulate human conversation. The introduction of ChatGPT on 30 November 2022 has shifted the way people approached AI, as it quickly became a popular tool in the academic community. In fact, as we are writing this review, a persona, Dr Emma Foster, powered by ChatGPT4o, supports the writing process. Chatbots can process user questions, respond with context-specific information or prompts, and adapt their feedback based on ongoing interactions. In educational contexts, chatbots act as digital mentors, responding to questions, providing hints, and adjusting their support based on user feedback. Chakraborty et al. [46] discussed diverse chatbot systems, including several designed for medical science applications.

But be careful, they do not always get things right, so human intelligence remains important! Chatbots and large language models (LLMs), such as ChatGPT, have become increasingly integrated into educational settings for their ability to generate instant explanations, assist in writing, and simulate interactive learning. However, despite their linguistic fluency and surface-level coherence, these systems frequently lack contextual understanding, domain-specific reasoning, and ethical awareness—all essential qualities in higher education and scientific training. One critical flaw of LLMs is their tendency to produce hallucinations (this phenomenon is known as AI hallucination), outputs that are grammatically sound and persuasive but factually inaccurate or entirely fabricated [47].

For example, in biological sciences, an LLM may confidently generate a plausible sounding but incorrect citation, misidentify a protein function, or suggest an unsupported experimental conclusion. A student relying uncritically on such output might unknowingly propagate scientific errors, misapply statistical methods, or reference non-existent studies. Without human intervention, these errors could go unnoticed and misinform assessments, lab reports, or even research articles, ultimately undermining the rigor and reliability of science education.

Moreover, AI systems are trained on vast datasets that often reflect systemic biases and gaps in representation. In healthcare and life sciences, where data predominantly come from Western or high-income populations, AI-based diagnostic tools may perform poorly on underrepresented groups. A prominent example is provided by Obermeyer et al. [48], who found that a widely used healthcare algorithm underestimated the medical needs of Black patients. The model used healthcare spending as a proxy for health status, which systematically disadvantaged patients from communities that historically receive less care. As a result, fewer Black patients were flagged for necessary follow-up care—reducing their inclusion by more than half. When the researchers revised the algorithm to base predictions on actual health outcomes rather than spending, this disparity was significantly mitigated.

Similarly, Chen et al. [49] emphasized that AI systems have the dual capacity to amplify or reduce disparities depending on how they are designed and implemented. In their review, they showed how AI can be used to identify underserved populations, create fairer prediction models, and monitor health inequities, such as the disproportionately high maternal morbidity rates among Black women in the United States. However, these benefits can only be realized when AI is developed with diverse training data, transparent evaluation criteria, and active human oversight.

Besides issues of factual accuracy and fairness, there is growing concern about the perceived legitimacy of algorithmic decisions. According to Shulner-Tal et al. [50], users are generally more skeptical of AI-driven decisions, especially in high-stakes or ambiguous scenarios, unless clear and understandable explanations accompany the decisions. Their study found that the type of explanation matters: people were more accepting of AI decisions when explanations emphasized sensitivity analysis (how changes in input data affect output) or certification (expert endorsement or validation). These findings reinforced the importance of explainable AI (XAI) in educational and scientific contexts, where understanding why a conclusion was reached is often more valuable than the conclusion itself.

In this light, the work of Mehrabi et al. [51] provides a foundational framework for identifying and mitigating bias in AI systems. Their survey outlines how algorithmic unfairness arises not only from biased data but also from poor algorithmic design, unexamined assumptions, and misaligned goals. They argue for a taxonomy of fairness approaches and highlight the indispensable role of human reviewers in auditing AI systems, especially in fields like natural language processing, healthcare, and education.

Together, these examples show that AI is not infallible, nor is it ethically neutral. Human judgment must continuously evaluate its outputs, especially in biological sciences, where context, nuance, and ethical considerations are critical. Educators must train students not only to use AI tools but also to question them—scrutinizing the logic behind their outputs, checking sources, and considering the broader implications of relying on automated systems. This dual approach—leveraging AI while maintaining human critical oversight—is essential to ensure scientific rigor, promote equity, and foster responsible innovation in the age of intelligent machines. As institutions increasingly adopt AI tools, the potential for promoting CT through personalized, interactive experiences becomes more pronounced. AI systems allow for real-time adjustments to learning content, fostering higher engagement and helping students practice analytical reasoning. By aligning academic material with individual needs and infusing prompt-based questioning, these tools encourage students to articulate and refine their thought processes. In so doing, they not only enrich learners’ educational journeys but also prepare them for the challenges of a technologically evolving environment.

The Role of Educational Chatbots in Biological Sciences

Educational chatbots hold great promise in the biological sciences because they can simulate interactive dialogues comparable to those found in small-group discussions or mentoring sessions [46]. For instance, a chatbot integrated into a genetics course might guide students through the analysis of DNA sequencing data, prompting them to explain their reasoning at each step. When learners propose an alternative hypothesis or misinterpret a strand alignment, the chatbot can offer scaffolded hints that encourage deeper investigation of underlying molecular principles. This conversational element helps students articulate their own thinking, refine it progressively, and reach more precise conclusions.

Let us consider another example from an ecology course. Here, chatbots can prompt learners to model interactions among species within a virtual ecosystem. Students receive diagnostic questions about population growth rates or the impact of resource depletion, and they respond by selecting or typing specific variables. The chatbot then evaluates their responses, adapting follow-up questions to clarify the relevant ecological theories. More advanced systems integrate knowledge bases containing curated biological facts, so that the interactions can include current real-world research trends. At each juncture, students confront new scenarios requiring analytical thinking and application of learned concepts. This iterative approach of inquiry and feedback reinforces conceptual connections and fosters a stronger grasp of scientific methods.

Another great use of a chatbot is practical lab settings. Here, chatbots may guide learners in setting up protocols for cell culture or polymerase chain reaction (PCR). Once students input their plan, the chatbot may highlight areas where reagent concentrations might need adjustment and ask clarifying questions about the purpose of specific buffers or enzymes. This blend of immediate guidance and active recall may contribute to improved problem-solving skills and greater confidence in understanding complex procedures [52]. Chatbots also reduce pressure for quick responses, creating an environment where students can reflect carefully before proceeding.

Overall, chatbots in biological education can extend opportunities for evidence-based reasoning, strengthen student engagement, and encourage consistent reinforcement of CT skills. Whether integrated into online learning platforms or embedded in adaptive tutoring systems, these AI-driven features create personalized dialogues that adapt to learners’ unique needs, ultimately enriching the learning journey in lab-based and theoretical biological curricula. In the following sections, we delve deeper into AI and other tools that facilitate CT.

3.3. Collaborative Problem-Solving with Artificial Intelligence Education

Research in AIEd is interdisciplinary and, as such, rooted in theories from various fields. One of the main theories identified in research papers includes the constructivist learning theory [53]. This is not surprising, as collaborative problem-solving with AI in education has roots in social constructivism, a theory suggesting that meaningful learning arises when students actively construct knowledge through interaction and shared reflection [54,55]. This forms the basis of collaborative-based approaches to teaching, such as inquiry-based projects, problem-based learning (PBL), and collaborative online environments, such as virtual labs [56,57].

Several studies have addressed collaborative problem-solving with the use of AI [58]. Let us explore some examples. In STEM fields, particularly in biological sciences, AI-powered platforms support this process by providing flexible environments where learners can work on authentic tasks. PBL is one such approach: small groups of students tackle real cases in genetics, ecology, or biomedical research, refining their hypotheses as they engage with AI-driven datasets and simulations. As they collaborate, students practice decision-making, interpret AI-generated suggestions, and discuss the rationale behind each step. Collaboration may also continue asynchronously, through the use of interactive platforms that promote co-creation of knowledge [59].

Another promising setting involves virtual labs, such as Labster (https://www.labster.com/). Here, AI tools simulate complex experimental scenarios in various areas of biological sciences, ranging from basic biology, microbiology, or protein structure analysis. Learners can manipulate variables within simulated lab protocols—often guided by AI-based feedback that flags improbable methods or unexpected results. While carrying out virtual experiments, student teams exchange findings, propose new directions, and collectively assess outcomes. This promotes higher-order CT and a collaborative ethos that parallels real-life research teams [58]. Educators facilitate these experiences by structuring reflective checkpoints where learners articulate their reasoning and evaluate AI-generated insights. Through these iterative dialogues and hands-on virtual scenarios, participants gain a deeper understanding of biological principles, sharpen their analytical skills, and cultivate a mindset geared toward continuous problem-solving.

4. Leveraging Artificial Intelligence to Enhance Critical Thinking in the Biological Sciences

CT, a cornerstone of postsecondary education, is crucial for the effective evaluation of information and decision-making in both academic and real-world scenarios. Defined as the ability to use data and evidence to make informed choices about what to trust and act upon, CT is highly valued by employers, who associate these skills with employees capable of making consistent, evidence-based decisions in their careers [60]. As the digital landscape evolves, the proliferation of tools like Google, which provide immediate access to vast amounts of information, has sparked debates about their influence on our reasoning capabilities and learning capacity.

Ultimately, the goal is to prepare students to harness AI technologies effectively while maintaining the integrity and rigor of scientific research. This requires a balanced approach that emphasizes the importance of human oversight and critical analysis in validating findings and preventing biases inherent in automated systems. By fostering a culture of CT, higher education institutions can equip biological scientists with the skills needed to navigate the complexities of modern research and contribute to the advancement of knowledge in an ethical and responsible manner.

4.1. Artificial Intelligence Is Not Just the Chatbots

AI encompasses far more than the chatbots we commonly interact with; its definition, while elusive, generally captures the capability of machines to emulate the intellectual faculties of higher organisms. An ideal AI system would not only possess self-awareness and logical reasoning abilities but would also have the capacity to gain knowledge through experience and respond dynamically to its external environment [1].

AI is not a monolithic technology but rather comprises various subfields, such as machine learning (ML) and deep learning, each adding layers of intelligence to applications either individually or in combination. Machine learning involves the study of algorithms that enable computer programs to automatically improve through experience, encompassing categories like supervised, unsupervised, and reinforcement learning, with ongoing expansions into semi-supervised, self-supervised, and multi-instance ML [61].

Recently, the concept of transformative AI (TAI) has garnered attention, highlighting AI’s potential to influence significant societal changes even without mimicking human-level cognitive abilities. This discussion extends into the realm of general-purpose technologies (GPTs), where AI’s transformative potential continues to be a focal point in scholarly literature [62].

In the biological sciences, AI’s application is vast, with innovations that drastically reduce the time and cost of experiments—tasks that traditionally took years. Advances in AI have revolutionized data generation and analysis, facilitating the efficient handling of diverse data formats, which is pivotal in academic research and the biotechnology industry. For example, the ability of AI to precisely identify the 3D structures of biological molecules, like proteins, plays a critical role in biological research [63]. An example is AlphaFold, which accurately predicts protein structures down to the atomic level, even when no similar structures are known [64]. AI further enhances innovation not just in laboratory settings but also across the entire lifecycle of pharmaceuticals and biochemical products, optimizing bioprocesses for enzyme production at pilot scales [65].

AI’s integration is essential for overcoming human cognitive limitations in data collection, integration, and hypothesis testing across various subdisciplines of biology. It stands as a cross-cutting technology that enhances our capacity to undertake comprehensive biological research.

In the educational sphere, AIEd is an emerging field with lagging applications compared to sectors like applied science and finance. As previously discussed, AI-enhanced education strives to provide personalized learning experiences, adapt content to individual needs, and foster active engagement in the learning process. According to Chen et al. [66], future research should not only focus on AI itself but also on the innovations that facilitate the development of technologies with human-like intelligence capabilities in decision-making, learning adaptability, and cognitive functions. Establishing a thriving ecosystem of innovators and advanced infrastructures is paramount for the successful integration of AI in education and beyond.

4.2. AI Applications in Biological Sciences Education: Enhancing Critical Thinking Through Technology

As described above, AIEd has blossomed into a robust field characterized by its diverse applications and theoretical approaches [37]. In the realm of biological sciences education, AI is not just revolutionizing traditional teaching methodologies but is also expanding the horizons of what can be achieved in both teaching and learning. This section explores the growing influence of AI in biology education, detailing its impacts on pedagogical approaches and providing examples of its current applications.

In science education, AI’s role extends beyond just tutoring. It has been effectively integrated as a tool to support teachers in roles such as facilitators, guidance counselors, evaluators, and providers of customized learning support. Moreover, AI-driven applications, including chatbots, are now common features in educational settings, offering tailored learning experiences and immediate feedback. This instant feedback is crucial, as it allows for the dynamic assessment of students’ progress by analyzing their learning patterns and adapting the educational content accordingly.

4.3. AI-Enhanced Laboratory Simulations

Science laboratories have long been foundational to science education. With the rise of computer-based learning, it is natural that this progression includes the development of laboratory simulations. Such simulations in science education can enhance both procedural knowledge, which is crucial for performing laboratory procedures, and conceptual knowledge, necessary for understanding and explaining what is demonstrated [67].

Recognizing the critical role of modern technologies in the educational process, various interactive science laboratories (ISLs) have been designed to support students in hands-on science experiments. Interactive teaching has become essential for fostering conceptual learning because it engages students more deeply by involving them directly in the learning process. The evolution of computing technology introduces new modalities in education, sparking greater interest among students by allowing them to utilize innovative tools and boosting their motivation to learn. In this context, virtual labs (VLs) and computer simulation-based labs (CSLs) emerge as cutting-edge solutions to challenges in science education where physical alternatives are unavailable, too costly, or excessively hazardous. Examples of ISLs are presented in

Table 3. Examples of interactive science laboratories (ISLs) in biology education.

Type of Laboratory	Examples	Description
Virtual Biology Labs	VHEalthlab https://www.vhealthlab.eu/, accessed on 1 May 2025.	An e-learning platform that allows students to simulated basic biology experiments, e.g., microscopy, cell division, PCR, and lab safety, among others.
Virtual Dissection Tools	Froguts https://thesciencebank.org/pages/froguts, accessed on 1 May 2025.	Digital platforms that allow students to perform virtual dissections, providing an ethical, cost-effective, and accessible method to study animal anatomy.
Molecular Biology Simulations	BioInteractive Simulations https://www.biointeractive.org/, accessed on 1 May 2025.	Tools that let students manipulate molecular structures and observe molecular interactions, aiding in the understanding of complex biochemical processes.
Ecological Simulations	SimUText Ecology https://simbio.com/simutext/, accessed on 1 May 2025.	Simulate ecological processes, such as energy flow, population dynamics, and evolutionary changes, to help students grasp ecological interactions and impacts.
Cell Biology and Microscopy	CellCraft https://cellcraft.io/, accessed on 1 May 2025.	An interactive game that teaches students about cell structure, function, and processes by allowing them to build and manage a virtual cell.
Genetics and Genomics	Learn.Genetics (University of Utah) https://learn.genetics.utah.edu/, accessed on 1 May 2025.	Virtual labs focusing on genetics, where students can conduct experiments like DNA extraction and gel electrophoresis, enhancing understanding of genetic science.

.

Importantly, ISLs play a significant role in developing CT among biological science students through a number of mechanisms. Below are the ways in which these virtual labs enhance CT skills:

• Problem solving and hypothesis testing. ISLs enable students to participate in experimental procedures in which they are required to use CT in order to solve problems or test hypotheses. For instance, through molecular biology simulations, students can experiment with DNA sequences to see what happens, and they are required to use logical thinking and predictive abilities to decide the outcome of their experiments.
• Application of theoretical knowledge. Virtual labs require applying theoretical knowledge to practical scenarios. In virtual dissection software or cell biology simulations like CellCraft, students apply their understanding of anatomy or cellular mechanisms to navigate and engage in the virtual environment effectively. This application of knowledge enhances understanding and retention.
• Decision-making under constraints. The majority of ISLs provide scenarios with variables that the students must control under constraints that mimic real scientific investigation settings. For instance, in ecological simulations like SimUText Ecology, students must identify the best way to control an ecosystem depending on a number of environmental and biological variables, thereby enhancing their ability to make decisions based on complex data.
• Development of inquiry-based skills. ISLs typically come with ‘what if’ features, through which students are able to inquire, a basic skill to attain inquiry-based learning. For instance, in VHEalthLab experiments, the students are provided with a branched case, giving them the option to choose different paths. Students may modify the parameters, execute the simulation, and observe other results, thus attaining a scientific mindset of querying, exploring, and understanding biological systems in meticulous detail.
• Enhanced observing and analyzing ability. In virtual labs, students must take note carefully and conclude from their experimentation results. As an example, genetics and genomics labs push the students to interpret genetic data and conclude their observations. Analytical and observational skills are practiced through the process.
• Feedback and reflection. All ISLs provide immediate feedback based on students’ behavior, a crucial aspect in reflective learning. The feedback enables students to learn what they did wrong or right in a safe environment and gain from the mistakes without repercussions in real life.
• Collaboration and communication. Certain ISLs are designed for collaborative activities, where students need to negotiate and communicate with other people in order to accomplish tasks or resolve issues. This exercise promotes CT because students must articulate reasons, consider other individuals’ perspectives, and negotiate solutions.
• Integration of cross-disciplinary knowledge. ISLs tend to bring together ideas from various scientific fields, which helps students synthesize ideas across disciplines. For instance, a lab could incorporate aspects of chemistry, biology, and physics, and ask students to synthesize these ideas critically to grasp a biological process comprehensively.
• Promoting creativity. Finally, by allowing a supportive climate to experiment and find out, ISLs encourage creative approaches to scientific research. Students can experiment with unorthodox procedures or come up with new ideas to problems without the risk of costly failures, thereby cultivating imagination and critical analysis.

5. Challenges Associated with AI

5.1. Using AI in Education: A SWOT Analysis

While AI tools have become integral to advanced biological research, many educational pathways do not dedicate enough time to elucidating how these systems reach conclusions. Students may be adept at using AI platforms but may lack a solid understanding of algorithms, training data, and model limitations. This gap can cause a superficial reliance on AI outputs without thorough evaluation of their accuracy or relevance [63]. Consequently, curricula in higher education must focus on enhancing digital literacy and analytical judgment. Incorporating coursework on data biases, interpretability, and model optimization enables students to move beyond simply operating software tools and toward critically appraising computational decisions.

This debate extends to advanced AI technologies, such as ChatGPT. While ChatGPT is renowned for generating coherent and persuasive text, it also poses challenges due to its potential to produce misleading or incorrect information, as its algorithms may rely on flawed data [68]. The significant implications for educational integrity raise concerns that AI might compromise academic honesty, misrepresent scientific facts, and encourage plagiarism [66]. In this context, it is crucial for educational programs to harness AI technologies in ways that enhance CT and learning, rather than detract from these foundational skills.

Ethical and responsible implementation is perhaps the key. In addition to technical training, AI education must address ethical imperatives, especially in biology. Data privacy poses significant concerns when processing sensitive genetic or medical records [69]. Likewise, biases in diagnostic models can amplify health disparities if developers overlook certain populations during dataset assembly. Future professionals in this field need to understand how prejudiced outcomes manifest within algorithms and exercise caution in applying those conclusions to real-world conditions. Classroom discussions and interactive case studies can further illustrate the structural factors that lead to algorithmic inequities, emphasizing the importance of ethical review at every stage of research.

Preparing future scientists for evolving responsibilities should be considered with care. Next-generation scientists and innovators will encounter increasingly powerful AI systems that extend into many facets of the biological sciences, including personalized medicine, climate change research, and genomic therapies. They will be tasked not only with leveraging these technologies but also with upholding the scientific integrity and ethical standards associated with them [69]. This responsibility includes communicating AI-generated results to peers and broader audiences in a transparent manner. Academic programs should cultivate these competencies by emphasizing collaborative projects that integrate computational approaches with research ethics. By developing a comprehensive, practical foundation, future experts will be positioned to guide the evolution of AI in ways that benefit both science and society.

As AI continues to permeate various aspects of scientific society, its integration into educational frameworks presents both promising opportunities and notable challenges. In higher education, particularly within the biological sciences, AI has immense potential for transforming learning environments and developing pedagogical strategies. The following SWOT (strengths, weaknesses, opportunities, and threats) analysis (Figure 2) explores how AI could influence the development of CT skills among students. By examining the strengths, weaknesses, opportunities, and threats associated with AI in education, we aim to provide a balanced perspective on its potential impacts. This analysis highlights the capabilities and enhancements that AI brings but also addresses the complexities and risks educators and institutions must navigate to effectively foster CT in an increasingly digital learning landscape.

As illustrated in Figure 2, while the strengths and opportunities illustrate how AI can enhance learning through personalized instruction, real-time data analysis, and greater accessibility, the figure also brings attention to underdiscussed but critical concerns. Under weaknesses, the figure points to issues such as AI’s inability to reason contextually, risks of student overreliance, and the generation of outputs without clear accountability or source verification. These align with growing concerns in the literature regarding the limitations of large language models and the potential for hallucinated or biased outputs, particularly when used uncritically in scientific contexts.

The threats quadrant emphasizes broader risks, such as the erosion of academic integrity, the spread of misinformation, and the potential loss of human mentorship in AI-mediated learning environments. These threats are particularly pressing in the context of biological sciences, where scientific accuracy, methodological rigor, and ethical oversight are paramount. As AI tools become more integrated into educational practice, there is a real danger that they may be adopted without sufficient safeguards, reducing opportunities for students to engage in reflective reasoning and peer-to-peer learning.

The inclusion of this SWOT analysis underscores our review’s central argument: that intentional curricular design, robust critical thinking pedagogy, and ethical oversight must accompany the effective use of AI in education. Instead of replacing human educators or critical judgment, let us position AI as a tool that prompts dialogue, supports inquiry, and enhances, rather than bypasses, student reasoning.

5.2. Human Criticality and Ethical Oversight in the Use of AI in Higher Education

AI technologies integrated into tertiary education hold immense promise, but they also present significant risks. While such tools promote efficiency in data management, personalized learning, and student support, they can be problematic if adopted uncritically. It is in this context that the labor of human criticality—defined as the reflective and evaluative act of interrogating information, methods, ethics, and implications—becomes vital. As foreshadowed in our introduction, the societal disruptions caused by climate change, the COVID-19 pandemic, and the rise of post-truth rhetoric have exposed the dangers of misinformation, algorithmic overreach, and data misinterpretation. During the pandemic, for example, AI-powered algorithms were used to track virus transmission, recommend resource allocation, and model public health scenarios. However, many of these tools demonstrated variable performance due to data gaps, model assumptions, and a lack of transparency, ultimately contributing to misinformation and public confusion [70,71]. This underscores a broader reality: AI is not inherently reliable or objective—its effectiveness relies heavily on human scrutiny.

This issue is particularly important in the context of biological sciences education. Students increasingly use AI tools—such as chatbots, bioinformatics programs, and automated data visualization platforms—to support learning and research. While these tools can be beneficial, they may also produce outputs that are incomplete, biased, or entirely fabricated, especially when generated by large language models prone to “AI hallucination” [47]. Without adequate training in critical evaluation, students may place undue trust in AI-generated content, potentially leading to flawed experimental design, misinterpretation of data, or the spread of misinformation.

As previously discussed, algorithmic bias is another area that necessitates human intervention. While AI has the potential to reduce disparities, it can also reinforce them when training data lack diversity or when ethical considerations are not built into the design. These challenges underscore the necessity of embedding ethical reasoning and fairness analysis into the use of AI—not only in healthcare and research but also in scientific education.

It is important to remember that AI is not an unbiased tool. Its training data, algorithmic logic, and deployment strategies inevitably reflect the assumptions and limitations of its human creators. As such, ethical oversight must be considered foundational—not optional. Core questions, such as:

What assumptions does this AI model make?

Whose perspectives or data might be missing?

What are the ethical implications of using this technology in a real-world biological or medical context?

…must be central to any educational curriculum that integrates AI.

Educators play a vital role in this process. Teaching CT cannot be limited to mastering scientific content—it must also include the development of ethical judgment, the analysis of algorithmic transparency, and an understanding of data provenance. As Shulner-Tal et al. [50] have shown, people are more likely to perceive AI decisions as fair and legitimate when those decisions are accompanied by explanations that are domain-specific and contextually relevant. This finding implies that AI systems in education should not only deliver information but also stimulate critical reflection and debate.

As discussed previously, Mehrabi et al. [51] advocate for the development of fair, interpretable, and accountable AI systems through active human–machine collaboration. Their comprehensive review of bias and fairness in machine learning further reinforces the need for continuous human involvement in identifying discriminatory patterns, challenging algorithmic assumptions, and promoting equity—particularly in high-stakes fields, such as education and healthcare.

Ultimately, AI should not replace human judgment in education but rather enhance it. A well-rounded curriculum that incorporates ethical literacy, critical data analysis, and reflective skepticism alongside technical competence will empower students to use AI tools responsibly and insightfully. In doing so, institutions can prepare a generation of biologists and scientists who are not only technologically adept but also ethically conscious—equipped to address the complex challenges of a data-driven world.

6. Future Directions and Recommendations

Before discussing potential areas for research and practice, it is important to acknowledge the limitations of this review. Herein, we relied on a narrative approach, drawing on a broad but not an exhaustive systematic set of sources. Because narrative reviews do not follow a formal protocol with clearly defined inclusion or exclusion criteria, there is a potential for selection bias and incomplete coverage of available literature. Additionally, rapid advancements in AI could have led to the emergence of new tools or research findings that appeared after our search window. Keeping this in mind, the following recommendations and future directions are proposed to guide educational institutions, researchers, and practitioners in effectively integrating AI to promote critical thinking in the biological sciences.

As we continue to witness rapid advancements in AI, it is imperative that educational institutions adapt their curricula to include these technologies in ways that critically engage students rather than merely streamline information delivery. This involves integrating AI tools and methodologies into the learning process to enhance students’ CT skills. By doing so, educators can create a more interactive and responsive learning environment, where students are encouraged to question assumptions, analyze data critically, and develop a deeper understanding of complex biological systems.

Moreover, the use of AI in education should be designed to complement and enhance traditional teaching methods, not replace them. This ensures that students benefit from the best of both worlds: the efficiency and personalization offered by AI, and the human touch and critical oversight provided by educators. For instance, AI can provide personalized feedback and adaptive learning paths, allowing students to progress at their own pace while receiving targeted support where needed. Educators can focus on fostering discussions, encouraging collaborative projects, and guiding students in ethical considerations and the critical evaluation of AI-generated outputs.

Below, we outline several areas where AI can be leveraged to promote CT in biological sciences. These can serve as guidelines to help institutions ensure that AI functions as a powerful ally in fostering CT and deep learning. This strategic approach not only prepares students to excel academically but also equips them with essential skills to address ethical and practical challenges they will encounter as future scientists in a technologically advanced world.

• Enhanced data analysis tools: Create AI-driven data analysis software tools that allow students to interact with real biological datasets. These tools can help students learn to identify patterns, test hypotheses, and draw conclusions, thereby enhancing their analytical skills.
• Simulated environments and virtual labs: Develop AI-powered simulations that model biological processes in the real world and laboratory experiments. Online science labs should give students realistic problems to solve using data, so they learn about biology without the limitations of real labs.
• Adaptive learning systems: Implement AI systems that adapt to the learning style and speed of each student, providing personalized pathways through course content. This has the potential to fill personal knowledge gaps and improve understanding of difficult concepts by allowing students to explore areas they are weak in, with AI-powered guidance to help them navigate tricky subjects.
• Collaborative AI projects: Implement project-based learning, whereby students utilize AI tools to tackle real-world challenges. These can be projects such as designing an experiment, analyzing the results using AI tools, or even coming up with novel AI applications. Team projects also can instill in students the importance of teamwork and communication, an essential in the modern interdisciplinary fields of science.
• Critical evaluation of AI outputs: Teach students to critically evaluate AI output, considering accuracy, ethical implications, and potential biases inherent in AI models. Class discussions, targeted coursework on AI ethics, and classroom activities exposing students to the limits and potential abuses of AI technology can accomplish this.
• Integration of AI in traditional curricula: Integrate AI tools with traditional pedagogy seamlessly. While AI can provide efficient and tailored learning experiences, human guidance is required for handling complex ethical issues as well as for providing the deep, nuanced understanding that comes from traditional education methods, like seminars and hands-on labs.
• Professional development for educators: Offer ongoing teacher professional development on emerging AI tools and pedagogical strategies. Educators need to be acquainted with these technologies to integrate them substantively into their instruction and to guide students in the effective use of these tools.

7. Conclusions

There has never been a more favorable time to rethink how we teach and cultivate CT in biological sciences. While AI offers significant advantages in streamlining teaching and assessment processes, it is crucial that we also maintain and nurture direct human mentorship. Human interaction remains essential for addressing individual student needs, fostering social interactions, and providing ethical oversight [44]. This balance between automated instruction and human guidance is part of a broader dialogue that affects educational strategies across disciplines.

Incorporating AI should not mean diminishing the role of educators but rather enhancing their ability to guide and inspire students. Educators play an irreplaceable role in interpreting AI outputs, contextualizing knowledge, and integrating ethical considerations into the curriculum. Their insight and experience are invaluable, particularly for navigating complex moral landscapes and tailoring learning experiences to diverse student populations.

Well-structured educational strategies and continuous engagement with intellectually challenging scenarios equip students and professionals with CT skills to address the complex problems of today’s rapidly evolving scientific landscape. As the field of biological sciences continues to advance, driven by technological innovations and interdisciplinary approaches, the importance of CT escalates. It guarantees that scientific progress is not only innovative but also responsible, considering the wider effects on society.

Ultimately, the goal is to create a learning environment where AI complements traditional teaching methods, thereby enriching the educational experience and preparing students for future challenges. By integrating AI tools thoughtfully, we can enhance the educational landscape, ensuring that students not only learn to use technology effectively, but also develop the CT skills necessary to use it wisely. This balanced approach will prepare the next generation of scientists to drive innovation and maintain high ethical standards in their work, contributing positively to society and the field of science.