Trust in the Machine: Between the Lab Bench and the Algorithm
Brianna Perales
Instructor’s Introduction
Students in the WR 152: Digital Worlds class examine and discuss issues related to technology and society. For their first major assignment, they work on writing research-based Exploratory essays. Unlike thesis-driven, argumentative essays, the open-ended Exploratory essay provides the freedom and the opportunity for students to explore their chosen topic without having to prove an argument or to come to a conclusion. The assignment encourages students to write an Exploratory essay, starting with a personal experience or observation that prompts them to ask an open-ended question, which they then examine in greater detail based on their observations as well as on outside research. I also encourage the students to include at least 1 photograph, table, graph, figure, or drawing in the essay. Although the essay starts with a personal experience or observation, it expands to consider larger issues and the implications of those issues. The format encourages creativity as well as builds research skills and lends itself to probing a difficult problem. The essay may or may not arrive at a clear-cut solution to the problem, but it always yields insight of some kind.
In her Exploratory essay “Trust in the Machine: Between the Lab Bench and the Algorithm,” Brianna Perales used her own experience as a student researcher in a campus research lab as a jumping off point to explore the use of AI in medical research. Her essay grapples with the idea of entrusting AI to assist with research that has far-reaching implications for human health. She writes that “finding a balance between innovation and caution” is a struggle in the lab context and talks about the importance of the need to “keep the human element of decision-making central”. Brianna’s essay is a great example of an Exploratory essay because she examines, from a personal lens, an issue where there are no easy solutions or answers. In a context where AI is being entrusted with more and more tasks, Brianna writes that “AI in medicine isn’t just a technical question. It’s a human one. It’s about trust between patients and doctors, between researchers and regulators, and between humans and the technologies we create.” Through the process of exploration, she comes to the realization that there is “not a single answer, but an ongoing conversation that researchers, doctors, patients, and policymakers will all have to take part in.” The essay, in some sense, reveals that the process of searching for a human-focused solution, is, by itself, the point of the exploration.
Malavika Shetty
From the Writer
The emergence of artificial intelligence (AI) has made technology play a critical role in the modern age of medicine as both a promising tool and as a source of uncertainty. As some researchers choose to develop their work by adopting AI use, others remain cautious of AI missing the mark in terms of reliability and ethical responsibility. The divide between the two stances has raised important questions about how and when AI should be integrated into medical device development and research.
This exploratory essay investigates these questions through the perspective of an undergraduate researcher working in a tissue microfabrication lab, where AI was not commonly used. Through the examination of this research environment compared to others that embraced the use of AI, a lab-to-lab debate can be extended into the broader medical field. Additionally, this essay gives space to look into earlier innovations in medicine, highlighting recurring tensions between technical innovation and caution. Rather than arguing for one side, this essay explores the complexity of AI’s role in medicine, emphasizing that technological progress is not only a scientific challenge, but also a human one.
Trust in the Machine: Between the Lab Bench and the Algorithm
Every Tuesday morning, I knew what to expect for the day. Most PhD candidates and lab members were ready by 10AM before the wave of meetings had begun. With various subgroups and programs, our lab meetings usually spilled over their allotted time and ended with us rushing to the next room either a floor above or below ours. Despite leaving only a few hours in the day to work in the wet lab, these group meetings were vital for lab members to communicate their progress on projects and pitch research ideas to one another for feedback or concerns with the current methods used in an experiment. As an undergraduate research student, these meetings served as a way for me to keep up with the different fields within the lab as well as our collaborators’ projects that involved our work.
Imaging what the future of medicine looks like with tissue engineering was easier as a member of the lab. The CELL-MET Program believes in the collective efforts of researchers across three universities to one day develop artificial tissue that can be used on the heart. Currently, there are several studies being conducted in smaller areas within the larger goal, which include facilitating the best protocols to carry out experiments that mimic the environment of the heart, studies using animal hearts, and studies involving cardiac microfluidics. The heart is one of our most complex organs, so we can undeniably expect some degree of error in experiments, but by how much? In this question, I found interest in the ways to minimize error and optimize the processes used, to get closer to one day fabricating lab-grown heart tissue for clinical and laboratory use.
During the summer, I spent my time making patterned stamps with a lab-made substrate that would host heart cells in efforts to create a constant, reproducible test environment for different drugs and diseases. When meeting with the other program members, some would relay stories about the use of artificial intelligence (AI) in their research since it made for an easier time throughout the program, especially since we would have to present at the end of the summer. It was then that I realized the use of AI within my lab was almost non-existent, if not actively discouraged. Instead, we heavily relied on outsourcing previously published research from other universities or just pure collaboration with other members of the lab. With cardiac function as a nuanced topic of study, many of my steps along the way were influenced by my mentor or outside advice from other senior members of the lab.
Why was AI embraced in some labs but avoided in ours? One researcher I spoke with had an AI system to help analyze imaging data, cutting hours of work into minutes. For me, my data was mainly qualitative which is simpler to do in person than it would be to put my imaging into an AI service. At first I felt like I was at a disadvantage, but I also wondered whether avoiding AI forced me to engage with the data more closely. Would I trust an algorithm to catch subtleties in cardiac function that even humans argue about? I began to see this question reflected on a broader scale in medicine itself. Surgeons, for example, face a similar dilemma. AI can process visual data in ways that outperform the human eye, but there’s hesitation in handling full control of something so delicate to a program.1 The more I thought about it, the more my own experience felt like a small echo of medicine’s larger uncertainty.
In a CELL-MET meeting, I was introduced to the concept of the organ-on-chip, a microfluidic device lined with human cells to mimic the physiological functions of a specific organ. Take a lung-on-a-chip device, for example. With this device, breathing motions can be replicated. As of now, these devices are used mainly to test responses to stimuli in vitro, but there have been major developments in medical devices such as these in which micro or nanochips are implanted into the human body. Fascinating developments like these in the medical field could potentially change patient healthcare indefinitely, but the path to getting advanced and reliable medical devices isn’t smooth sailing.
Due to uncontrollable factors like biological rhythms, physiological conditions, or functional integrity, the creation of a “perfect” medical implant device becomes a harder goal to reach day by day. However, the use of data algorithms from known or observable health factors can be utilized for a device backed by countless amounts of data. In many cases, this reliability has a backbone made from artificial intelligence.2In just a click of a button, AI programs can retrieve data from all over the world and send it to a computer, a laptop, or even a phone in seconds. AI can also provide suggestions on questions a user inputs, and, shockingly, its answers have become more and more accurate each time it is used. It has come to a point where it may be time to examine its use and the potential benefits it could have in different fields of work. When it comes to the medical field, however, its use may be scrutinized as many people see issues with dependence on AI for our own health or patient privacy.
While my peers had more of a way to use AI to assist them in their research, I remember sitting at my desk for hours on some days. I would read through various papers, trying to figure out how I could keep up with my fellow lab members who have had years of experience backing them. With so much ground knowledge to cover, it almost seemed impossible to be able to keep my head afloat when beginning my research project. I couldn’t just plug any code into ChatGPT or tell it to do my part of the work for me. Physically needing to be in the lab to make my stamps
and image them made me wonder if I was behind with the pacing of my experiments since optimizing my time wasn’t something AI could directly do. Although there could have been a scenario in which I would have struggled less by using AI, I not only grew more concerned with relying on it in the context of my research but also for professionals in the health field. If I had used it alone, I would feel more paranoid about the underlying issues that I might not have seen had I just accepted the word of AI. Those who are years into their work may be more cautious of utilizing AI, but there could still be a chance that a potential error caused by AI is laying underneath the cracks. When looking into these considerations, it’s vital that we look back to where integration of technology in medicine started.
Before the incorporation of AI in the medical field, nanotechnology had already made its mark as an up-and-coming resource for creating more effective treatments. Some of these nanotech devices include highly targeted drug delivery with the use of nanomedicine or enhanced diagnostics and the use of nanoscale imaging probes to provide early and accurate detection of disease. We also see the use of nanomaterials to expand the field of regenerative medicine, used for organ regeneration or tissue repair by promoting cell growth. Aside from these new aids in patient and health care, nanotechnology does have its drawbacks. As of now, there is not much known about the effects of nanotechnology which could include internal factors like biocompatibility and possible toxicity from nanomaterials within the human body.3
When I first heard about nanotechnology, I remember thinking it sounded futuristic, almost like science fiction. Now, the same thing is happening with AI. Both are fields that sit at the edge of excitement and unease. The optimism comes from imagining the breakthroughs. Diseases detected earlier than ever before, surgeries made safer, devices that adapt to a patient’s unique biology. But on the other side are the concerns. Toxicity in nanotechnology and bias and regulation in AI. Regulators in particular face a difficult task. Unlike a pacemaker or an insulin pump which perform their functions consistently, an AI algorithm has the ability to evolve. If it is constantly updating itself, what does it mean to approve of it as a device? Regulators have suggested that a system-level approach might be needed; one that accounts for the fluid nature of AI while still protecting patient safety.4 The balance of innovation and caution keeps AI from being fully embraced in some labs, including my own.
The issue of bias also cannot be ignored. AI systems learn from data, but the data we give them reflects the world as it already is, with all its imperfections and inequities. If most of the training data comes from one demographic group, the device may perform worse for everyone else. This isn’t just theoretical; there have been reports of AI in dermatology failing to properly assess conditions in patients with darker skin tones.5 In the context of medical devices, where the stakes are literally life and death, this kind of bias is detrimental to the treatment of different kinds of patients. It raises the question of whether AI will ever be objective enough to handle patient care without reinforcing existing disparities? Or is bias something we will always have to manage, no matter how advanced technology becomes?
Another layer is the ethical one. Even if an AI device is accurate and regulated, will patients accept it? Public perception plays a major role in the adoption of medical technologies. Microchip implants have become a recent topic of discussion. Even though implantable devices have been around for decades, like pacemakers, cochlear implants, and insulin pumps, the idea of a chip linked to AI raises deeper questions about autonomy, privacy, and identity. Studies exploring public perception show that acceptance often depends on not just what the device can do, but how people feel about the idea of technology inside their bodies.6 Some embrace it as progress while others see it as invasive. I noticed this when I talked with my family members about the scope of the research I was doing. Older members of my family voiced discomfort with the idea of “computers in their body” while mainly my younger relatives thought the idea of it was cool. For the older generation, these new devices seem like something they would have seen out of a movie in their youth. For the younger generation, it seems like a true possibility that humans could become “part-computer” even if that technically isn’t the case. Nonetheless, the slightest hesitation in accepting these advanced technologies can reflect how public perception can slow down the adoption of even the most promising devices.
That tension between enthusiasm and unease is visible everywhere in discussions of AI. At the American Medical Association, for example, there has been a growing push to encourage “responsible integration” of AI in healthcare. On the one hand, AI is praised for its ability to improve efficiency and expand access to care. On the other hand, physicians are warned not to rely on it blindly and to keep the human element of decision-making central.7 This duality reminds me of my own summer. In one space, AI was a helpful collaborator; in another, it was something to be avoided. Neither side was entirely wrong, but together they painted a picture of a field in transition.
By the end of the summer, I didn’t feel like I had a firm answer about whether AI should or should not be used in medical devices. Instead, I had a clearer sense of why the debate is so complicated. I understood why my lab discouraged it, and why the uncertainties about regulation, ethics, and bias made it safer to stick to traditional methods. I also saw why other labs embraced it, because the potential benefits were too great to ignore. Optimized large data analysis, generated code for specific needs, and more can give us the answers we’re looking for, potentially saving us years of research, but, also, saving the lives of others at a faster rate. What stayed with me most was the realization that AI in medicine isn’t just a technical question. It’s a human one. It’s about trust between patients and doctors, between researchers and regulators, and between humans and the technologies we create.
As I look ahead, I know these questions will only become more pressing. AI is not going away. Medical devices will continue to get smarter, nanotechnology will continue to evolve, and the public will continue to wrestle with what it all means. My summer didn’t give me a definitive stance, but it gave me a way of thinking about the issue. I now see AI not as a solution or a threat, but as a tool whose value depends on how carefully we use it, and whose risks depend on how honestly we confront them. For me, that is the real implication of AI in advancing medical devices. Not a single answer, but an ongoing conversation that researchers, doctors, patients, and policymakers will all have to take part in.
Notes
1. Hashimoto, Daniel A., Guy Rosman, Daniela Rus, and Ozanan R. Meireles. “Artificial Intelligence in Surgery: Promises and Perils.”, 7-8. Annals of Surgery 268, no. 1 (July 2018): 70–76. Accessed September 25, 2025. https://pmc.ncbi.nlm.nih.gov/articles/PMC5995666/.
2. Ibrahim, Hamza Khalifa. “From Nanotech to AI: The Cutting-Edge Technologies Shaping the Future of Medicine.” American Journal of Applied Sciences (Special Issue from the First Libyan Conference on Technology and Innovation) 2024, 414. Accessed September 25, 2025. https://aaasjournals.com/index.php/ajapas/article/view/810/723.
3. Ibrahim, 412-413.
4. Gerke, Sara, Boris Babic, Theodoros Evgeniou, and I. Glenn Cohen. “The Need for a System View to Regulate Artificial Intelligence/Machine Learning-Based Software as Medical Device.” NPJ Digital Medicine 3, no. 53 (2020). Accessed September 25, 2025. https://www.nature.com/articles/s41746-020-0262-2.
5. Obermeyer, Ziad, Brian Powers, Christine Vogeli, and Sendhil Mullainathan. “Dissecting Racial Bias in an Algorithm Used to Manage the Health of Populations.” Science 366, no. 6464 (2019): 447–453. Accessed September 25, 2025. https://www.science.org/doi/10.1126/science.aax2342.
6. Minnich, Carlin. Unveiling the Invisible: A Societal Odyssey into Microchip Implants and Public Perspectives. PhD diss., Walden University, 2021. Accessed September 25, 2025. https://www.proquest.com/docview/3233914900?pq-origsite=gscholar&fromopenview=tr ue&sourcetype=Dissertations%20&%20Theses.
7. Robeznieks, Andis. “Big Majority of Doctors See Upsides to Using Health Care AI.” American Medical Association, January 12, 2024. Accessed September 25, 2025. https://www.ama-assn.org/practice-management/digital-health/big-majority-doctors-see-u psides-using-health-care-ai.
Works Cited
Gerke, Sara, Boris Babic, Theodoros Evgeniou, and I. Glenn Cohen. “The Need for a System View to Regulate Artificial Intelligence/Machine Learning-Based Software as Medical Device.” NPJ Digital Medicine 3, no. 53 (2020). Accessed September 25, 2025. https://www.nature.com/articles/s41746-020-0262-2.
Hashimoto, Daniel A., Guy Rosman, Daniela Rus, and Ozanan R. Meireles. “Artificial Intelligence in Surgery: Promises and Perils.” Annals of Surgery 268, no. 1 (July 2018): 70–76. Accessed September 25, 2025. https://pmc.ncbi.nlm.nih.gov/articles/PMC5995666/.
Ibrahim, Hamza Khalifa. “From Nanotech to AI: The Cutting-Edge Technologies Shaping the Future of Medicine.” American Journal of Applied Sciences (Special Issue from the First Libyan Conference on Technology and Innovation), 2024. Accessed September 25, 2025. https://aaasjournals.com/index.php/ajapas/article/view/810/723.
Minnich, Carlin. Unveiling the Invisible: A Societal Odyssey into Microchip Implants and Public Perspectives. PhD diss., Walden University, 2021. Accessed September 25, 2025. https://www.proquest.com/docview/3233914900?pq-origsite=gscholar&fromopenview=tr ue&sourcetype=Dissertations%20&%20Theses.
Obermeyer, Ziad, Brian Powers, Christine Vogeli, and Sendhil Mullainathan. “Dissecting Racial Bias in an Algorithm Used to Manage the Health of Populations.” Science 366, no. 6464 (2019): 447–453. Accessed September 25, 2025. https://www.science.org/doi/10.1126/science.aax2342.
Robeznieks, Andis. “Big Majority of Doctors See Upsides to Using Health Care AI.” American Medical Association, January 12, 2024. Accessed September 25, 2025. https://www.ama-assn.org/practice-management/digital-health/big-majority-doctors-see-u psides-using-health-care-ai.
Brianna Perales is a sophomore studying Electrical Engineering in the College of Engineering. From her experience in technical design and biomedical research, she has formed a strong interest in the influence of emerging technologies on medical innovation within the past few years. In her free time, she is an active member of Terrier Motorsport and partakes in on-campus research in the field of cardiac tissue engineering. Her experience in these activities have contributed greatly to kickstarting her interest in the intersection of engineering and medicine.
She extends her thanks to her WR 152 professor, Malavika Shetty, for her support and guidance throughout the course, which helped elevate her writing skills and broadened her range as a writer.