In Part One, Chapter Three, a new protagonist arrives in the lovely family home in Newmarket, a town close to the Alps. It is a pigeon, and there is a reason I decided to introduce this bird to the list of actors in my book.
The pigeon Galina is necessary for the first book as, thanks to her, I will introduce concepts such as LLM hallucinations and object detection. She will play an even more critical role in the second volume, which I will write in between September 2025 and November 2026. While the first book is focused on quantum computing and the ways to optimize neural networks, the second story will be about AI in medicine.
Humans should never forget that the world has plenty of intelligent creatures who don’t belong to the homo sapiens category. Researchers trained pigeons to peck at cancerous breast tissue pictures using associative learning. According to the paper “Pigeons (Columbia Livia) as Trainable Observers of Pathology and Radiology Breast Cancer Images,” 1 the pigeons spent a few days training on how to visually differentiate between cancerous and noncancerous tissue. The birds were given a set of brand-new images of breast tissue to diagnose. They were capable of identifying cancerous tissue 85 percent of the time. The accuracy levels jumped to 99 percent when the responses of several birds were pooled. These pigeons scored higher than the human radiologists who were given the same task.
Like humans, pigeons have the visual ability and perceptual machinery to notice the difference in detail between cancerous and benign tissue and the cognitive power to place these two types of tissue into separate conceptual categories. Apparently, human cognition isn’t superior to these birds’ born-in capabilities when spotting cancerous tissue.
Why should we be surprised when AI is able to provide support in diagnosing breast cancer? Human radiologists should be overjoyed when they have software that frees their time to focus on patient communication and human care. Beyond saving time, AI might be better than us at certain tasks. After all, human eyes might not be the best visual tool to serve a patient.
In 2014, I went through a painful experience in a hospital in Bonn. During a standard check-up, there was a suspicion that I might have a lump. As my family history isn’t great when it comes to breast cancer, the doctors decided to operate. There was nothing there, but I spent a day in the hospital, needed to undergo anaesthesia, and could not run in the morning for several weeks while my wound healed. If AI software had been involved in analyzing my mammography, I might have avoided the intervention.
I’m looking forward to hearing your stories about how AI changed—or could have changed—your day-to-day experiences in healthcare, education, or picking the suitable ingredients for an apple pie.
- Richard M. Levenson, Elizabeth A. Krupinski, Victor M. Navarro, and Edward A. Wasserman, PLoS ONE 10, no. ↩︎