As machines are increasingly being trained to take over human tasks, how will pathology continue to retain its human approach? What does a more "human" pathology profession actually look like? And could AI help to restore professional autonomy in a burnt-out workforce?

Inspired by an interactive session at USCAP 2026, led by Proscia, and David West’s subsequent blog, The Pathologist decided to put these questions out for discussion. We assembled a panel including an AI developer, digital pathology directors from the USA and the UK, and an early adopter of AI in the lab.

In part one, the panel discusses where AI development priorities should lie in order to move toward a more human practice.

What does a “more human" pathology profession actually look like?

“Embracing AI isn't just about adopting new technology. Pathologists also need to evolve the role they play.”

Syed T. Hoda: I'm a strong advocate for cultural change within pathology. My sense is that pathology has traditionally been a quiet specialty. That's always struck me because, by nature, I'm a loud person in a quiet profession.

I have great respect for my colleagues, and I understand why the culture has developed that way. But I think the profession now needs to redefine how it sees itself within healthcare. That means becoming more visible, more engaged, and more willing to advocate for the value that pathologists bring. To me, that's an important part of making pathology more human.

I think we're approaching a fork in the road. There's some truth in the idea that if you embrace AI, you're more likely to thrive alongside it. But embracing AI isn't just about adopting new technology. Pathologists also need to evolve the role they play. If all we do is use AI to perform the same tasks more efficiently, without becoming more collaborative or contributing in new ways, the profession risks becoming less communicative and less connected to the rest of healthcare.

That's why leadership matters. Pathologists need to work more closely with colleagues across healthcare, as well as with technology companies and the pharmaceutical industry, to shape how AI is implemented. Used thoughtfully, AI can be a catalyst not just for technological change, but for a broader cultural transformation that makes pathology a more visible, connected, and ultimately more human profession.

“Pathologists can spend more time applying their expertise where it has the greatest impact.”

David West: I completely agree. We've already seen what embracing AI looks like in software development. AI has fundamentally changed the way engineers write code. There was an understandable period of uncertainty as developers worked out what these new tools meant for their profession, but today many developers find that AI makes coding more productive, more creative and, frankly, more enjoyable. It hasn't eliminated the need for software engineers; in many cases, it's created more demand for them, and it's changed the nature of their work.

I think we're going to see a similar evolution in other knowledge-based professions. That means we need to think carefully about what success looks like for pathology in that context. My hope is that technology companies and pathologists work together to shape that future, rather than allowing it to happen around them.

If AI can reduce the time pathologists spend on routine reporting and administrative tasks, it creates the opportunity for them to spend more time collaborating with clinicians, contributing to treatment decisions, and applying their expertise where it has the greatest impact. To me, that's a much more exciting vision than simply making today's workflows incrementally more efficient.

“AI should address the humanity of our workforce.”

David Gibbs: When we talk about making pathology more human, we have to start with the patient.

Ultimately, the greatest benefit AI can deliver is faster, more accurate access to the right diagnosis, enabling patients to receive the most appropriate treatment as quickly as possible. As precision medicine continues to evolve, diagnostics are becoming an increasingly integral part of treatment decisions. In that sense, improving pathology is one of the most direct ways we can improve patient care.

But we also need to think about the humanity of our workforce.

If AI is introduced simply to extract more productivity from pathologists and reinforce a production-line mentality, we'll risk making the profession even less satisfying. At a time when we're already facing significant workforce shortages, that's the last thing we should be doing. Unless pathology remains an attractive career, those shortages are only going to become more severe.

I agree with David that we should be using AI to take over repetitive, low-value tasks. At the same time, digital pathology also creates opportunities to improve working lives in other ways. It offers greater flexibility through remote reporting, makes it easier to participate in multidisciplinary team meetings, and gives pathologists better access to specialist expertise. That's particularly important for rare subspecialties, where expertise may be concentrated in only a handful of centers.

For me, a more human pathology profession improves the experience of both patients and pathologists. If the AI conversation becomes dominated by cost savings and workforce pressures, we'll have missed the bigger opportunity.

“AI can recreate some of the benefits of working within a multidisciplinary team.”

Derek Welch: I see a more human pathology profession as one where technology makes information easier to access and interactions more natural.

We're all becoming accustomed to tools like ChatGPT and Claude, where you can ask questions in plain language and receive immediate, useful answers. I think AI can bring that same kind of conversational experience into pathology.

For more than a century, pathologists have often had to work with limited clinical information. A specimen arrives with only the details provided on the request form, and you're left trying to interpret the case without knowing what you don't know. AI has the potential to change that by bringing together the patient's clinical history, previous pathology, imaging, laboratory results, and treatment history into a single, searchable interface that can be queried in real time.

The same principle applies to the slide itself. Rather than simply viewing an image, a pathologist could ask natural language questions such as, "Do these lymph nodes contain metastatic disease?" or "If so, where are the suspicious areas?" That creates an experience that's much closer to discussing a case with a colleague.

That also has important implications for quality. Diagnostic errors sometimes occur because pathologists don't have access to all the relevant clinical context. By intelligently collating information from across the patient's journey, and making it readily available, AI can reduce uncertainty, avoid unnecessary work-ups, and help pathologists make better-informed decisions.

To me, that's the human parallel. AI can recreate some of the benefits of working within a multidisciplinary team – bringing together the collective knowledge surrounding a patient and making it available at the moment it's needed, even if that collaboration is happening virtually rather than face to face.

Are the AI tools being developed for pathology solving the right problems?

“The biggest pain points are often the administrative burdens that take time away from specialist work.”

David West: Pathology AI has become largely synonymous with diagnostic algorithms because they make compelling headlines. Those capabilities are certainly interesting, but the latest generation of AI – particularly transformer-based models, including large language and vision-language models – is opening up opportunities to address much broader challenges.

When we speak to pathology laboratories, the biggest pain points are often the administrative burdens that take time away from specialist work. These include report generation, case assignment, and ancillary orders, not just activities within the pathologist's diagnostic workflow but across the laboratory as a whole.

I believe the real opportunity for AI to solve problems is by clearing this administrative load. Ultimately, pathologists are highly trained specialists whose expertise is most valuable when they're working at the top of their licence.

“AI has much greater potential in improving the practical aspects of laboratory work.”

Syed T. Hoda: I would like to say yes, but I think the answer is mostly no.

Tumor detection was an interesting and relatively straightforward problem for AI to tackle, but I don't think it has been a game changer in routine practice. It’s a routine diagnosis with a black and white answer, for which AI has largely provided a shortcut rather than introducing fundamentally new capabilities.

Where AI has much greater potential is in improving the practical aspects of laboratory work. The routine work that David mentioned – case assignment, workflow optimization, quality assurance, and quality control – could benefit enormously. These are opportunities for AI to add genuinely new value without interfering with the pathologist's diagnostic role, making the technology more broadly useful across laboratories.

“AI must address a clearly defined problem.”

David Gibbs: The important point is that AI must address a clearly defined problem. If you don't understand the question you're trying to solve before implementing a tool, you risk creating additional work for pathologists rather than reducing it.

Beyond image analysis, there are many other opportunities. Earlier today I was discussing AI systems designed to automate incident reporting, manage laboratory processes, and support tasks such as reagent management. There is a huge amount of innovation happening in these areas.

However, if you look at the current priorities for NHS England, for example, the immediate focus remains on digital pathology and digital image analysis. Those technologies are seen as the foundation on which wider adoption of AI will be built.

To what extent do you think efficiency gains should drive AI development in pathology?

“Efficiency shouldn't be the sole objective for AI or digital pathology because efficiency alone doesn't inspire people.”

Syed T. Hoda: Looking at our own data, we've found that pathologists who work quickly on the microscope tend to remain quick on the digital system, while those who take longer still tend to take longer. Digital pathology certainly delivers efficiencies in other areas, particularly in logistics and workflow, but the impact on individual reporting speed is much less clear-cut.

More importantly, I think our emphasis on efficiency reflects a broader cultural issue within medicine. Clinicians are constantly encouraged to work faster, become more productive, and deliver more in less time. While those goals have their place, they can also make medicine feel less human.

Efficiency shouldn't be the sole objective for AI or digital pathology. If that's the only message, adoption will be limited because efficiency alone doesn't inspire people. Pathologists are driven by curiosity and the opportunity to solve complex diagnostic problems. The most compelling AI applications will be those that genuinely deepen understanding, enhance scientific practice, and enable new ways of thinking – not simply those that promise to make existing tasks a little faster.

“If digital pathology is introduced primarily as a way to increase productivity, it's understandable that there will be resistance.”

David Gibbs: We also need to recognize the sense of threat that can accompany these technologies.

If digital pathology is introduced primarily as a way to increase productivity – by asking pathologists to report more cases, monitoring their performance, or otherwise changing the way their work is assessed – it's understandable that there will be resistance. Those messages can make technology feel like something being done to pathologists rather than for them.

In practice, though, once pathologists begin working routinely with high-resolution digital displays, many find the experience more ergonomic and, in many respects, easier than using a microscope. They start to appreciate the benefits for themselves.

The key is to take people on that journey. Adoption doesn't happen overnight, and not everyone will be an early adopter. Some pathologists only become comfortable with digital pathology after they've seen trusted colleagues embrace it and demonstrate its value in everyday practice. That's a perfectly natural part of the transition.

“Better patient outcomes are what will ultimately determine the commercial success of these technologies.”

Derek Welch: For me, there's one overriding objective that matters above all others: improving quality for patients.

When I look at the AI tools currently being developed, their greatest opportunity is to close the small but important gap between what pathologists can achieve today and what we might consider the ground truth or the highest possible level of diagnostic accuracy. Pathologists are exceptionally accurate, but, like any professionals, they can make mistakes. AI has the potential to reduce that already low error rate across different disease areas, and that's where I believe it can have the greatest clinical impact.

Better patient outcomes are also what will ultimately determine the commercial success of these technologies. Payers are understandably most interested in tools that improve outcomes while reducing the overall burden of care, both in terms of patient health and healthcare costs. In my view, that's the strongest case for AI adoption.

Beyond diagnosis, we're also seeing a new generation of AI tools that provide prognostic insights. Rather than simply helping pathologists identify disease, these systems can predict outcomes such as the risk of recurrence or the likelihood of responding to therapy. We use a prostate cancer tool, for example, that generates prognostic information much more quickly and easily than traditional molecular techniques such as PCR-based assays or next-generation sequencing. As these capabilities mature, they have the potential to become an important part of routine pathology practice.

The secondary benefits, such as improving pathologist efficiency and turnaround times, are important, but they're fundamentally business drivers rather than the primary goal.

Newsletters

Receive the latest pathologist news, personalities, education, and career development – weekly to your inbox.

I have read and understand the Privacy Notice*

Subscribe