Traditionally, the dissemination of evidence-based knowledge is through information taught at medical schools, academic conferences, and peer-reviewed literature, including systematic reviews. Experts from other fields rarely teach in medical schools, and we seldom venture beyond health care literature for evidence-based information. There are few opportunities for the exchange of ideas between basic scientists and clinical researchers, which further delays the application of translational research to clinical practice. Medical training relies on years of education and learning from clinical experience. This process of building upon previous knowledge while constructing new knowledge is known as “constructivism.” Constructivism works well for established paradigms of knowledge generation at a controlled pace. However, in settings such as the pandemic and currently with GenAI, evidence is generated rapidly and often sourced from disciplines outside health care. In the example above, it was the interdisciplinary work of aerosol scientists that helped us acknowledge airborne transmission [3,12].

Constructivism takes time and is a gradual process. As individuals, we are unable to process numerous sources of information simultaneously, and technology can help address this challenge by personalizing the information we receive. When faced with new ideas that do not align with previous knowledge, we either ignore new information or work harder to reframe our understanding. Our psychological drive for consensus suppresses disagreement, sometimes even in the face of evidence-based facts [8,19]. Thus, when exposed to information contrary to previous knowledge, we fail to recognize our limitations and misunderstandings. When inundated with a flood of new research from untraditional sources, the challenge is even greater. With the acceleration of knowledge generation, it is impossible to stay informed with advances in literature. Hence, there is a need to belong to communities that amplify accurate information. Learning to distinguish misinformation from tried and tested evidence-based facts becomes an additional role for physicians. As a scientific community, we are at risk of overlooking errors and missing crucial discoveries, especially if we fail to listen to experts from other fields. We further distinguish between constructivism and connectivism in Table 1.

Public health institutions’ announcements from the World Health Organization and the Centers for Disease Control and Prevention reach billions of people who trust and follow their recommendations as they are based on well-founded evidence. Until fully verified, informing the public of aerosolized spread could cause panic and chaos, leading to a shortage of personal protective equipment with a huge socioeconomic impact [10]. They rely on data from multiple well-designed studies before conclusively deciding on airborne spread. Although the gold standard, this method is not completely foolproof and has its drawbacks. Many randomized controlled trials are refuted within a few years of publication, with a considerable number not published [20,21]. There are biases leading to higher publication rates of studies with positive results. Biomedical science has been in a silo, failing to learn from other disciplines and often discarding information that is obvious to other disciplines [13]. In fact, if the need for unanimity overrides motivation and authority, it can be detrimental, with negative consequences for true learning [22], and one way to combat this as a scientific community is to test and retest a hypothesis to maintain scientific rigor [23]. Overworked clinicians, stretched for time, often rely on online resources and opinions of local experts, practicing eminence-based medicine if there is insufficient evidence-based medicine. Over time, as more evidence emerges from multiple sources, the science becomes clearer, the truth emerges, and the findings are more universally adopted by physicians in practice. This process can take several years to get from the bench to the bedside, sometimes up to 5 years in surgical specialties [20].

Connectivism was first proposed by Siemens [24] as a learning theory for a digitally connected world. In connectivism, new knowledge resides in the diversity of opinion and is distributed across networks, and learning occurs when individuals form connections across these networks [24,25]. Knowledge in a digital world where information is changing rapidly resides in several sources, experts, and networks, serving as “nodes” of information [26]. New knowledge resides outside of individuals, in digital networks and online communities [25,27]. Learning is not about acquiring new information but connecting to networks of knowledge [26]. Thus, knowing “where” and “who” instead of “what” and “how” becomes the new paradigm for information acquisition and learning. Health care experts need to be active participants in this dynamic network, serving as “nodes” that can recognize accurate information while discarding misinformation. The internet has allowed for the creation of communities and knowledge networks; learning how to navigate the diversity of opinions, differentiate reliable sources of information, and distinguish accurate facts from misinformation is, therefore, a valuable skill for future physicians. The theoretical approach to connectivism relies on the diversity of opinions, the strength of connections, the ability to recognize patterns, and the notion that the capacity to know is more important than what is already known. [25,26,28]. Coherence provides synchronization of different elements in this virtual community and provides organization and structure in the diverse networks of knowledge [29].

We consider the following implications for using connectivism as clinicians: First, the starting point of connectivism are “nodes” of accurate information; these can be individual experts, devices, or communities. Knowledge is created through interaction between the nodes and due to the diversity of opinions [26,28,30,31] by maintaining and nurturing connections and forming alliances with experts across disciplines. In these knowledge networks, Connectivism supports lifelong learning, collaboration, and critical thinking and fosters the ability to synthesize information from multiple sources [28,32]. The diversity of nodes makes it possible for networks to adopt and evolve over time [26]. The 4 foundational aspects of successful networks in connectivism are autonomy, connectedness, diversity, and openness [33].

Second, autonomy allows learners to choose their learning communities through self-directed learning in massive online open courses. Individuals have the freedom to express their ideas and explore the knowledge network to identify experts and acquire knowledge based on their level of motivation and not be restricted by a hierarchical approach to learning. Exploring the diversity of opinions allows the learner to experience the breadth of opinions. A physician with the skills to identify reliable and accurate sources of information with a diversity of opinions has a better understanding of the challenges faced by patients and can provide practical advice by directly addressing misinformation. These dialogues allow for interactivity and problem-solving [29].

Connectedness enables collaboration, and learning occurs through new connections between nodes and being connected to the right sources allows knowledge propagation in networks [29]. Maintaining, nurturing, and contributing to this network with coherence allows for continual learning, with new knowledge residing in the community rather than the individual. Connection with the right people in the right context contributes to learning. The ability to find and analyze new information quickly is crucial. With rapid changes in scientific understanding, staying current by collaborating with experts outside one’s field is essential. Currency and staying current with accurate information is the intent of connectivist learning [34].

Learning in connectivism is based on the production of knowledge rather than the consumption of knowledge by prioritizing relevant data while filtering out irrelevant information. In the digital environment, learning is enhanced by engagement with virtual communities of practice through blogs, posts on X, podcasts, and meaningful dialogue [35]. Interactions are central in the online educational experience, and for deep learning to occur, these interactions need to be purposeful with systematic discourse [36]. With the recent interest in GenAI, we can recognize that knowledge resides outside of humans and in this increasingly connected world, solutions to problems are provided by teams from multiple disciplines. GenAI can be used to provide personalized learning through communities, fostering a greater sense of autonomy and motivation [28,37].

If academia fails to embrace a connectivist framework, it risks losing its position as an arbiter of information while becoming less relevant.

The application of connectivism helps expand our approach to constructing scientific knowledge. It can move us from traditional classrooms to virtual communities by implementing solutions that are responsive, agile, networked, and dynamic [26]. When faced with emergent challenges with the rapid growth of information from multiple disciplines, Connectivism can bolster a swift response. We envision several applications of connectivism through social learning platforms that already exist for students to solve complex problems using online open-access tutoring through self-motivation [38]. Community-driven GenAI solutions already exist connecting 5 million users to computing resources for learning with open access to the latest technologies [39,40] with the potential to offer personalized learning and cloud-based integrations of GenAI for smart classrooms [41,42]. These social learning approaches demonstrate that connectivism directs the learner to a socially shared source of knowledge. The idea of distributed cognitive theory proposed by Hutchins [43] is relevant to connectivism in that it provides the framework and approach to lifelong and self-determined knowledge acquisition in social learning contexts

A limitation of connectivism is the overwhelming amount of information and diversity of opinions. Educators should address this by organizing the content by highlighting the common themes through coherence [29]. Another solution is teaching learners to identify the credibility and cognitive authority of web-based resources [44,45]. Teachers can tailor the lesson to the individual needs of learners, and consistent use of internet-based tools across lessons improves the learner’s confidence and learning ability to use these tools [46]. We provide a practical guide for the application of connectivism in Table 2. We highlight the need for understanding before applying its principles in real-world contexts by creating feeds and recognizing sources of information. Another solution to information overload is the use of effective filters for feeds and autogenerated information. Education around using social media and digital scholarship is still being defined and will need refinement and guidance from educators, although connectivism, at its core, should be self-directed by the learner.

First, creating a dynamic learning network encourages health care professionals and students to build and maintain professional networks through platforms like LinkedIn, X, and specialty-specific forums. Individuals can access real-time updates and exchange information. Use the feed generators from online communities and interested studies to build a knowledge network. Demonstrate the use of desired feed from journals on topics of interest. Identify experts from various disciplines.

Second, integrate technology into learning and practice; demonstrate the use of available resources such as podcasts, online webinars, blogs, and community forums to help with knowledge-building.

Third, develop critical information literacy skills, including training on how to critically assess the reliability and credibility of information sources, especially those found online. It is essential for learners to be able to distinguish between valid and misleading information and address misinformation.

Fourth, by leveraging artificial intelligence capabilities, clinicians can filter out unnecessary information, and GenAI tools can identify relevant scientific articles through feeds from reliable sources of information and amplify their search using freemium tools such as Research Rabbit. In addition, these tools allow for adaptive personalized instruction to the learner.

Fifth, the clinician can incorporate these tools as part of a synchronous classroom activity and model critical inquiry skills when using GenAI. By analyzing the accuracy of outputs, clinicians can model iterative prompting to refine GenAI outputs, refine learning, and expose the gaps in outputs.