This content was originally published on The Conversation.
Historical AI Promises
Back in 1958, a groundbreaking roomsize computer known as the Perceptron was unveiled, featuring innovative circuitry. The U.S. Navy projected that this machine had the potential to develop humanlike qualities such as walking, talking, seeing, writing, reproducing, and even consciousness.
Fast forward to the present day, and we see echoes of these claims in today’s artificial intelligence landscape. Despite the passage of over sixty years, the similarities between past and current AI promises are striking.
Boom and Bust Cycles in AI
The field of artificial intelligence has experienced a repetitive cycle of success and setback since its inception. Throughout different periods of technological advancement, the promise of AI has often been met with subsequent disappointments. While optimism fuels progress, it is crucial to acknowledge and learn from the failures of the past.
Frank Rosenblatt, the inventor of the Perceptron, laid the groundwork for AI with his creation. This electronic analog computer was a learning machine designed to categorize images into two distinct groups. The Perceptron’s mechanism involved physical connections via wires, reminiscent of contemporary artificial neural networks found in popular AI models like ChatGPT and DALL-E.
Similar to modern machine learning processes, the Perceptron adjusted its connections following incorrect predictions to enhance future accuracy. Today, large language models such as ChatGPT can generate extensive text-based responses and link images to text inputs to create novel images based on given prompts. These AI systems continually improve through user interactions.
Rise and Fall of AI Hype
In the aftermath of Rosenblatt’s Perceptron, prominent figures like Marvin Minsky foresaw the development of machines with human-like intelligence by the late 1970s. Despite initial progress, achieving human-level intelligence remained elusive.
The inherent limitations of AI systems were highlighted by their lack of contextual knowledge, rendering them ineffective in resolving language ambiguities that humans effortlessly navigate. The initial AI “winter” set in during 1974, following setbacks attributed to the perceived shortcomings of the Perceptron.
However, by 1980, AI experienced a resurgence, marking the onset of the first official AI boom. The emergence of expert systems tailored to specific knowledge domains propelled AI back into the spotlight.