Can machines think?
That was the question Alan Turing asked in his 1950 paper that introduced the the concept that later became known as the Turing Test. But I’m getting ahead of myself.
Before we go deeper down the rabbit hole, it’s helpful to cover some basics about Artificial Intelligence or AI. Let me start by saying that I’m no AI expert. I’m coming at it as a layperson with zero background or training in computer science or any related field—approaching AI from the perspective of a curious learner. Over the last eighteen months, I’ve delved into AI more than the average person by logging dozens of hours reading and listening to experts talk about it on podcasts. Let’s start with a brief history.
A Brief History of AI
(The primary source for this history is Ethan Mollick’s book, Co-Intelligence.)
The fascination with machines that can think dates back to 1770 and a mechanical chessboard. It was set upon an elaborate cabinet whose pieces were manipulated by a robot dressed as an Ottoman wizard. Known as the Mechanical Turk, people were stunned after it beat such figures as Ben Franklin and Napolean at chess. The secret? A real chess master inside manipulating its fake gears. It was all a lie.
In 1950, a toy and a thought experiment changed people’s thoughts about artificial intelligence. Claude Shannon, inventor and information theorist, developed a mechanical mouse named Theseus that could navigate a maze. Around the same time, Alan Turing, also known as the “father of modern computing,” published a paper that laid out how machines might mimic intelligent humans human behavior. The paper began with the question, “Can machines think?” It later became known as The Turing Test.
The term “artificial intelligence” was coined in 1956 by MIT’s John McCarthy. Early successes in solving logic problems and playing checkers led researchers to predict AI would beat chess grandmasters in a decade. But a series of “AI Winters”—periods of stalled progress and funding—slowed those expectations
The current AI boom began in 2010, driven by machine learning techniques for data analysis and prediction. “Supervised learning” allowed companies to use labeled data—like facial recognition images paired with names—to make accurate predictions. This same technology powers the personalized ads you see after online searches.
Supervised learning had limitations, but in 2017, Google’s paper “Attention Is All You Need” transformed the AI landscape by introducing the “Transformer” architecture. This breakthrough allowed AI to focus on the most relevant parts of a text and understand language more like humans, paving the way for Large Language Models (LLMs) like ChatGPT.
And there you have it! A quick overview of the history of AI. Next week, we will pick up with some basic terminology you’ll need to understand the conversation about AI.
Questions to Consider…
What did you believe about AI before reading this? How has that changed?
If machines could think, what would be the ethical implications? Should they receive the same protections as humans?
How comfortable are you with the idea of AI making decisions on behalf of people? Why or why not?
