The Origins and Goals of Artificial Intelligence
The quest to create artificial intelligence has transformed from a niche academic pursuit into the central mission of the world’s most powerful technology companies. At Google’s headquarters, the ambition has shifted from simply organizing information to solving intelligence itself. This pursuit is driven by the belief that recreating human thought can help solve every other problem facing humanity. However, as machines perform tasks once thought to be exclusively human, we must ask whether our intelligence is a profound mystery or merely a sophisticated set of patterns.
The dream of creating a thinking machine became a formal scientific pursuit with the rise of digital computers. Early pioneers like Alan Turing and John von Neumann viewed computers as symbol manipulators that process information by pushing around 0s and 1s. This led to the founding of artificial intelligence as a field during a 1956 workshop at Dartmouth College. Organized by mathematician John McCarthy, the gathering brought together key figures who believed that every aspect of learning could be described precisely enough for a machine to simulate it.
The early decades of artificial intelligence revealed a fundamental split in how to achieve human-level intelligence. One camp focused on symbolic AI, which uses human-readable words and rules to represent knowledge and solve logic puzzles. A different approach, known as subsymbolic AI, sought inspiration from biology by creating simple computer programs modeled after brain cells. Psychologist Frank Rosenblatt created the perceptron, a system that learns through a process called supervised learning by adjusting its internal weights based on thousands of examples.
These two methods represent a divide between conscious logic and unconscious perception. While early researchers successfully programmed computers to beat grandmasters at chess, they struggled to replicate the flexible, intuitive thinking humans use every day. Tasks that a toddler finds easy, like understanding a conversation or navigating a cluttered room, proved incredibly difficult for machines. Today, while specific subsymbolic methods have become dominant, the field continues to grapple with whether intelligence is a matter of following the right rules or learning the right patterns.
Full summary available in the Readsome app
