Bi-Weekly Geopolitical Report – Distinguishing My Wife From a Hat, an AI Story (June 26, 2023)
Thomas Wash | PDF
In his book, The Man Who Mistook His Wife for a Hat and Other Clinical Tales, Oliver Sacks, a neurologist, details his experience with patients suffering from varying neurological disorders. In one such case, he dealt with a man who had lost the ability to recognize faces. The patient was a university music teacher who had always been known for his calm and collected demeanor but had suddenly began behaving strangely. He would sometimes fail to recognize his students and was often seen patting the top of water fountains and parking meters as if they were small children. His antics were widely regarded to be jokes since he didn’t have trouble communicating, and his musical ability was as good as it had ever been.
It wasn’t until the patient was diagnosed with diabetes that he sought professional help. Aware of the disease’s impact on his eyesight, he visited an ophthalmologist, who reassured him that his vision was fine but referred him to see Dr. Oliver Sacks for a neurological exam.
During the visit, Dr. Sacks noticed something was off about the patient. The man seemed to be having trouble perceiving Dr. Sacks fully. Instead of looking directly at Dr. Sacks’ face, the patient fixated on certain parts. He gazed at Dr. Sacks’ nose, chin, right ear, and right eye, but never his face as a whole. After telling the patient the exam was over, the man attempted to find his hat but instead reached for his wife’s head and tried to lift it as if he were about to put it on. To Dr. Sacks’ surprise, the patient’s wife treated this as if it were an everyday event.
The case of the man who mistook his wife for a hat is a great illustration of how artificial intelligence (AI) neural networks process information. Like the patient, neural networks do not have the ability to look at an entire image to judge what it is. Instead, they break down images into parts, specifically into an array of numbers called pixels. AI models (neural networks) see images by recognizing patterns in the numbers that represent the image. They can make distinctions between different objects through training which then teaches them to associate certain patterns with specific objects. Just like the patient who needed to see an eye, nose, and mouth to know that he was looking at a face, AI models need numbers to achieve the same task.
This report provides a beginner-friendly introduction on how AI learns and processes information. We will begin by discussing the similarities between AI and our brains. Next, we will explain how AI works and explore some of its most important applications. We will then discuss some of the challenges and limitations of AI. We end the report with market ramifications. While this is not intended to be an exhaustive summary of AI, readers should come away with a stronger understanding of the technology and why it is such a big deal.