As AI rapidly transforms medicine, pathology has often been described as one of the specialties most vulnerable to automation. Yet the reality inside diagnostic practice appears more complex. Reflecting on AI-assisted pathology, war technology, and human judgment, this essay explores why the future may demand not fewer pathologists, but stronger ones.
As the long holiday came to an end, both the temperature and humidity began to rise, reminding us that the hot season is approaching.
Many people still seemed to be away, and the trains were not as crowded as I had expected.
During the holidays, the only time I touched my PC was for less than an hour each day to write this blog. Other than that, I merely did a few searches on my smartphone.
The rapid progress of autonomous AI systems in the war in Ukraine has become a major topic, and people have already begun to imagine a future in which wars are fought mainly by robots. That future may arrive as early as next year.
The idea that attacks on people and infrastructure could determine the course of war is deeply unsettling.
Human ingenuity will likely focus on how to defend against such attacks, but even then, people will probably consult AI for strategy. In that sense, the situation feels completely contradictory.
People often discuss how AI may take away human jobs, but how much work will actually disappear because of AI?
Pathology was once considered one of the medical fields most likely to be replaced relatively early by AI, but it now seems there are clear limits to that prediction.
I myself sometimes consult AI to make sure there are no missing differential diagnoses, but up to that point, I still have to guide the process.
As image databases continue to accumulate, some diagnoses are certainly becoming easier, and there will probably be more situations where human specialists only need to confirm the results. Even so, a vast number of problems remain beyond what AI alone can solve.
AI-assisted diagnosis will undoubtedly become more useful in the future, but paradoxically, that future may require pathologists with even greater abilities.
There may no longer be much need for humans to handle tasks that beginner pathologists can already manage.
However, the reasoning behind a diagnosis and the way the diagnosis itself is constructed must still be led by the pathologist.
AI merely waits for instructions, as if asking, “So, what shall we do?”
To put it extremely, even if one places a specimen on the stage and says, “Diagnose this,” AI can do nothing if the specimen itself is incorrect.
Pathologists must possess enough skill to give AI appropriate instructions, and acquiring that skill naturally requires effort. In the end, what we must do may not be all that different from what was required when I first became a pathologist.
There is something profoundly valuable contained within what human beings have patiently built over generations.
The more powerful our tools become, the more deeply human wisdom itself may be tested.
・・・
