Artificial Intelligence Is Not a Calculator: A Sociotechnical Mediator of Knowledge and Action for Engineers,Scientists,and Health Professionals

As professionals—a nurse and two engineers—we love a good analogy. But we question the analogy that generative artificial intelligence (AI) through widely available large language models (also known as “AI”) is like a calculator. It is not. A calculator acts on data. In contrast, AI mediates the relationship of sharing data, information, knowledge, and wisdom (DIKW) among people. This commentary is intentionally coauthored by a nurse and two engineers to reflect the cross-disciplinary nature of AI’s impact.

We argue that AI should be treated as a sociotechnical system that reshapes how knowledge is produced, shared, and validated (Oerther and Oerther, 2026c).

Many professionals, including educators, engineers, nurses, physicians, and scientists, warn of the risks of widespread use of AI, including potential negative impacts on the development of important skills such as critical thinking among adolescents (Nagata et al., 2026); diminished critical thinking skills among college students (AACU, 2026); and negative impacts on scientific writing among clinical fellows and early-career medical faculty (Steiner, 2026). Many of our colleagues in nursing and engineering are sincerely worried that more AI use equates to less human thinking. But we also know of many colleagues who reject this traditionalist stance and promote the “correct” use of AI (Mohanasundari et al., 2023). Often these progressive colleagues argue from the logic that “AI is just the next tool.” Often, we hear these progressive colleagues ask as a rhetorical question, “if you use a calculator, should you not also use AI?” And we have heard explanations of why the calculator analogy fails (Bailey, 2025).

In our view, AI as a calculator is the right idea but the wrong analogy, and here is why.

Prior to 2020, many professionals enjoyed the benefits of face-to-face (also known as F2F) interaction at conferences—including regional, national, and international venues. Events including F2F activities facilitate deep interaction through oral presentations and poster discussions and most meaningfully through the layered, interpersonal relationships developed with peers as part of intense, intellectually stimulating side conversations. These engagements include the exchange of cognitive information as well as the exchange of metadata that comes from human contact. But conferences are expensive, often exclusive, and therefore may represent a barrier to a level playing field for scientific engagement.

Zoom changed that. But there was a cost.

Through online, digital platforms such as Zoom, FaceTime, Google Meet, MS Teams, and WebEx, professionals are now able to “hear”—and more importantly to “see”—colleagues around the globe for what seems to be near-zero marginal cost. When coupled with effective pedagogy, Zoom may be used to engage with students (Oerther and Peters, 2020a). Zoom allows global networks of professionals to chat more frequently than traditional F2F meetings that require significant physical travel. But we also know there is less metadata from a Zoom interaction, and we each have experienced the reality of “Zoom fatigue” (Bailenson, 2021). As professionals, we have come to balance the benefits of Zoom and the costs of F2F (e.g., balancing online with IRL—or in the real world—interactions). In education, the availability of online, digital platforms has supported the adoption of “hyflex” pedagogy—including the HyFlex Learning Community (e.g., https://www.hyflexlearning.org). For all of these reasons, we hold the view that Zoom mediates the relationship of sharing DIKW among people.

AI is not a calculator. Like Zoom, AI mediates the relationship of sharing DIKW among people. For environmental engineers, scientists, and health professionals, this mediation may increasingly shape how we interpret sensor data, design remediation, and treatment processes and forecast their performances; manage civil infrastructure assets; assess risk; communicate uncertainty to administrators, communities, and other stakeholders; or engage in community-scale health interventions while still preserving engineering judgment and domain expertise and the caring and trust that are hallmarks of the nursing profession.

The practical implications of AI as a sociotechnical mediator are best understood through the hierarchy of DIKW. As illustrated in Fig. 1, AI does not merely calculate values; it facilitates the transition from raw environmental or clinical data to actionable professional wisdom, such as identifying the need for stream reaeration or medical intervention.

OPEN IN VIEWER

FIG. 1.

AI as a Sociotechnical Mediator of Knowledge and Action. This conceptual framework illustrates AI not as a standalone “calculator” or tool but as a digital infrastructure that mediates the flow of data, information, knowledge, and wisdom (DIKW). On the left, multidomain inputs (infrastructure data, environmental sensors, and health indicators) flow into a central AI hub. On the right, this mediation facilitates collaborative engagement among diverse professionals—including nurses and engineers—to produce resilient, equitable, and sustainable engineered solutions that reflect caring and trust. Side-by-side examples highlight the application of this mediation in environmental engineering (stream dissolved oxygen levels for trout survival) and nursing (blood lead levels in children). AI, artificial intelligence.

The benefits of AI include rapid access to data and information (e.g., similar to a Google search) as well as synthesis similar to knowledge and wisdom (Too Long; Didn’t Read, TL;DR). But the costs of AI include hallucinations, bias, privacy issues, etc. AI must be regulated with good practice, including transparency, accountability, and verifiability (Oerther and Oerther, 2026a). This framing aligns with a growing body of work characterizing AI as a sociotechnical infrastructure rather than a standalone tool (van de Poel, 2020). Consistent with this framing, emerging literature suggests an exponential increase in AI adoption across environmental science and engineering, with over 65% of environmental tasks transitioning from conventional statistical models to AI-based approaches (Konya and Nematzadeh, 2024). We view this trend as a reflection about AI’s value and the risk of AI being inseparable from energy use, carbon footprint, and potential health impacts. This situation calls for a close collaboration among environmental engineers and scientists, health professionals, and the professionals responsible for the improvement and expansion of AI infrastructure.

There are also major concerns around potential environmental costs associated with AI (Crawford, 2024). These include energy consumption, water consumption, and carbon emissions, which are directly relevant to environmental engineers, scientists, and health professionals. These also include emerging threats to environmental health, including pollution from low earth orbit associated with the accidental—and intentional—decay of satellite systems (Pultarova, 2024). Those who want to “forbid” the use of AI are correct to note that there will be costs to humanity in adopting this new tool.

In our opinion, the “problem” with AI is not its use; rather, the problem is the unaccounted cost of transition (i.e., the cost to “re”educate everyone—from the pre-K student and teacher to the doctoral candidate and examination committee). And here is where the Zoom analogy fits perfectly in our view. The pivot from F2F instruction to Zoom instruction included a huge cost (Oerther and Peters, 2020b)! During April 2020, approximately one-half of the human population was under some form of “COVID lockdown.” In much of the United States, the majority of human interactions outside the immediate home pivoted to digital, online platforms, such as Zoom. This included K-12 education, higher education, religious services, playdates for children, etc. The result was a massive investment in the “re”education of human interaction (Oerther and Oerther, 2022). And now that society is on the other side of that investment, we may take for granted the massive effort that was involved.

As we incorporate AI into our daily lives, the question we raise is, “when is society going to make an investment in AI education on a level similar to the investment in Zoom? (i.e., when are we going to “shut down” and work exclusively on AI for six weeks?).”

Institutions of higher education are adopting AI policies. For example, The Ohio State University’s AI Fluency Initiative (The Ohio State University, 2025) and Purdue University’s “AI working competency” graduation requirement (Fiorini, 2025) aim to achieve comprehensive awareness and mastery of AI across all faculty and students. But who is going to pay for these skills?

Between 2020 and 2023, COVID cost the U.S. economy $14 trillion, with the largest contributors to costs including mandatory closure and avoidance behavior (Walmsley et al., 2023). But how much did the transition to Zoom increase productivity, facilitate remote work and hyflex education, and improve collaboration, including global collaboration? Although the use of Zoom is only a small portion of the total, from the mid-1990s to 2025, the U.S. digital economy grew from nonexistent to $4.9 trillion annually, or roughly 18% of the total gross domestic product (Deighton and Kornfeld, 2025). In a tech-driven future, two kinds of knowledge are essential, namely (1) digital literacy—the ability to use digital tools and understand how they function and (2) digital fluency—the ability to adapt, evaluate, apply, and share knowledge with others (Slagg, 2025).

In conclusion, our message is simple. The appropriate analogy for thinking about AI is not to debate “AI as a calculator.” Rather, it is essential to recognize that AI is a tool to mediate the relationship of DIKW among people, including how we model, monitor, and manage complex environmental systems. The gains to be expected will likely be huge. What society needs to urgently address is the question of how to pay for the “re”education?

We envision an education framework where next-generation engineers, scientists, and health professionals are systematically trained to effectively shape how AI is designed, taught, governed, and deployed, rather than merely adopting it. Such frameworks offer a pathway toward more resilient, equitable, and sustainable engineered systems. We view this as both an opportunity and a responsibility for the Environmental Engineering and Science and Health Professional community.

AI Statement

During the preparation of this work, the authors wrote a final version and then submitted the following prompt to an enterprise version of Google Gemini 3.1 Pro: “Assume the role of Professor Ramana Gadhamshetty, the editor in chief of Environmental Engineering Science (https://journals-sagepub-com-s.web.bisu.edu.cn/home/eena). Evaluate the attached article for inclusion in the journal. Be sure to consider Aims and Scope as well as Submission Guidelines, including quality and format of references. Before providing your final evaluation, perform a harsh self-critique considering hallucination, bias, tone, etc. Use the critique to improve your final evaluation. Before you begin your review, do you have any questions to ask or assumptions to clarify using a human-in-loop approach?” After using this service, the authors reviewed and edited the content as needed and take full responsibility for the content of the published article.

Author verified TL;DR = “AI isn’t a calculator—it’s more like Zoom, changing how we engage with knowledge. The real risk is not thinking less, but failing to invest in the education needed to use AI well.” (as recommended by Oerther and Oerther, 2026b).

Authors’ Contributions

D.B.O. conceptualized the study, developed the “Zoom” analogy, and wrote the original draft. S.O. and V.G. provided critical improvements to the sociotechnical framing and environmental engineering technical context, respectively. All authors reviewed, edited, and approved the final version of the article.