The History of Artificial Intelligence
From “Can machines think?” to machines that act.
You already use artificial intelligence every day. This chapter is not here to introduce it — it is here to help you understand it well enough to make smart decisions about it in your career.
Why History Matters for Business
For your generation, artificial intelligence is simply part of the furniture: built into your phone, your search bar, and the apps you open without thinking. So it is fair to ask why a business course would spend its first chapter looking backward.
The answer is that the kind of AI you grew up with is already changing. Early AI mostly answered — you asked a chatbot a question, it replied. The technology now moving into companies acts: it books the travel, reconciles the invoices, and manages multi-step tasks with limited supervision. The professionals who thrive will not be the ones who can use a chatbot — everyone can do that. They will be the ones who understand what these systems really are, where they fail, and what they cost.
That understanding does not come from headlines. It comes from the whole arc — which is what this chapter gives you.
The Foundations
Before there were computers powerful enough to be intelligent, there were the ideas — and a single famous question.
Alan Turing asks the question
The story begins with a British mathematician. In 1936, Alan Turing described a “universal machine” — an idea that became the theoretical foundation for every computer built since. During World War II he helped crack the German Enigma code, proving machines could do work once thought to need a human mind.
Then, in 1950, he asked a deceptively simple question: “Can machines think?” Rather than debate it, he proposed a practical test. If a person holding a text conversation cannot tell whether they are talking to a human or a machine, we should be willing to call the machine intelligent. This became the , and it still frames how we argue about systems like ChatGPT today.
The first programs and the first name
- In 1951–1952, the first real programs appeared — checkers programs that could play a full game, one of which could even improve with experience.
- In 1956, a workshop at Dartmouth College gave the field its name: artificial intelligence. This meeting is, in effect, AI’s birth certificate.
- In 1958, the Perceptron — an early trainable — planted the seed of an idea that would not fully bloom for fifty years.
The Turing Test is why we still argue about whether today’s chatbots are “really” intelligent — a question that shapes how much we trust them with real decisions.
Early Pioneers and the First Setback
The first burst of optimism produced dazzling demonstrations — and the first hard lesson in the gap between promise and reality.
After Dartmouth, researchers built ambitious programs: one proved mathematical theorems, another tried to solve a wide range of puzzles. But the most famous program of the era was ELIZA (1966), which imitated a therapist simply by reflecting your words back as questions.
ELIZA understood nothing — yet people poured out their feelings to it and insisted it understood them. That tendency to see intelligence where there is none is now called the ELIZA Effect, and it is just as relevant today, when a fluent chatbot can seem far wiser than it actually is.
The first AI winter
These programs made great demos but could not deliver on the sweeping promises researchers had made. Computers were too slow, and the methods did not scale. As reality fell short of the hype, funders pulled their money. This collapse — the first — is one of the most useful episodes in the whole story, because the pattern of overpromise, disappointment, and retreat has repeated more than once.
The ELIZA Effect is a warning every manager needs: a system that sounds convincing is not necessarily one that is reliable. Judge AI by results, not by how human it feels.
Expert Systems and the Rise of Machine Learning
After the winter, a better idea took hold: instead of telling the machine every rule, show it examples and let it learn.
Expert systems: bottling expertise
The 1980s tried to capture human experts as thousands of hand-written rules. These had real successes in chemistry, medicine, and business — one saved a computer company tens of millions of dollars. But they were brittle and expensive: every rule had to be written by hand, and they could not learn. As they collapsed under their own maintenance costs, a second AI winter set in.
A different approach
The escape was a fundamentally different idea: let systems learn the rules from data. This is . As the internet generated oceans of data, machine learning found its first mass-market uses — the search engines, shopping recommendations, and streaming suggestions you still use today. For the first time, AI was quietly making money at scale.
This is the era when AI stopped being a science experiment and became a profit engine. The recommendation systems behind modern retail and streaming were the first proof that AI could drive real revenue.
The Deep Learning Revolution
Three forces finally arrived together — and AI leapt from useful to astonishing.
The public got a preview when, in 1997, a computer beat the world chess champion (through sheer brute-force calculation, not learning). But the real turning point came in 2012, when a system crushed the field at a major image-recognition contest and convinced researchers this was the way forward. In 2016, another system mastered the game of Go — a game that rewards intuition, not just calculation. And in 2017, the arrived: the design behind virtually every major AI system since, including the ones you use now.
Why did this happen now?
It is tempting to credit one clever algorithm, but the breakthrough needed three things at once: internet-scale data to learn from, better algorithms to give models the right structure, and powerful to do the training. The company whose chips proved ideal for this became one of the most valuable in the world — a thread we pick up in the next module.
“Why now?” is the question every business leader should ask of any technology. The answer here — data plus hardware plus algorithms — explains why AI exploded in the 2010s and not the 1980s, and why access to data and computing power is now a competitive advantage.
Generative AI Goes Mainstream
For the first time, AI was not hidden behind the scenes — it was a tool ordinary people used directly, every day.
The transformer unlocked a new kind of system: the , trained on vast amounts of text to generate fluent language. Between 2018 and 2020, these models revealed a startling principle — simply making them larger and feeding them more data produced genuinely new abilities.
Then, in November 2022, a simple chat box put that power in everyone’s hands. It reached an estimated one million users in five days and one hundred million within two months, becoming the fastest-growing consumer application in history. A race quickly followed among the major technology companies, each releasing its own assistant.
This is the moment AI became something every employee and customer could touch. Almost overnight, “Can we use AI for this?” became a question in nearly every department of every company.
The Agentic Era
The chatbot era was about AI that answers. The era you are entering is about AI that acts.
Two developments drove the shift. First, learned to work through problems step by step instead of answering instantly, making them far more reliable. Second, those models were connected to real tools — calendars, code, databases, the web — and given the ability to take action. The result is the AI agent: a system that can carry out a multi-step goal like researching a topic, drafting and sending a report, or reconciling a set of accounts.
Adoption is real
This is no longer a laboratory curiosity. Business adoption of AI has climbed sharply — one widely cited university report found organizational use jumped from 55% to 78% in a single year, and by its next edition reported adoption near 88%, with four in five university students using generative AI. Banking and finance, which this course examines in a later module, are among the fastest adopters.
The honest counterweight
But this course is not a sales pitch, and the history teaches caution. Alongside genuine progress sits a sobering reality: most organizations investing heavily in AI still struggle to turn it into measurable results, and many ambitious projects are scaled back or cancelled. The value is real, but it is concentrated among those who deploy it wisely.
You are entering the workforce at the exact moment AI shifts from answering to acting. The newest chapter echoes the oldest pattern in this story — a real breakthrough arriving alongside real limits. The skill that matters most is the judgment to tell them apart.
AI and the Jobs You’re Headed Toward
The history you just read is not only about technology — it is about the job market you are about to enter. Every shift in this story changed what work looked like, and the current one is no exception.
The big picture: more jobs, but different ones
The most-cited global study of this question, the World Economic Forum’s Future of Jobs Report 2025, surveyed over 1,000 employers representing more than 14 million workers. Its headline finding is more encouraging than the doom you may have heard: it projects roughly 170 million new jobs created and 92 million displaced by 2030 — a net gain, but with enormous churn underneath. The same report found that 86% of employers expect AI to transform their business by 2030, and that about 39% of the average worker’s core skills will change over five years.
Two kinds of opportunity
For a business student, AI creates two distinct paths, and you do not have to be a programmer for either.
- New AI-specific roles. The fastest-growing jobs in percentage terms are technology roles — AI and machine-learning specialists, big-data specialists, and fintech engineers lead the list. The U.S. Bureau of Labor Statistics projects data-scientist roles growing about 34% between 2024 and 2034, far faster than the average occupation.
- Existing business roles, reshaped by AI. This is where most business graduates will actually feel the change. Marketing analysts use AI to personalize campaigns; finance teams use it to detect fraud and model risk; HR uses it to screen and support employees; operations teams use it to forecast demand. The job title stays familiar — the toolkit changes. Each later module in this course explores one of these areas.
The skills that keep their value
Here is the part worried students tend to miss. The same research that tracks AI’s rise also finds that the most sought-after skills remain deeply human: analytical thinking tops the list, followed by resilience and flexibility, creative thinking, and curiosity and lifelong learning. AI handles the routine; judgment, communication, and the ability to ask the right questions become more valuable, not less.
You do not need to become an engineer to thrive alongside AI. The winning combination for a business professional is fluency with AI tools plus the human skills — judgment, creativity, communication — that AI cannot replace. This course is designed to build exactly that fluency.
The History of AI at a Glance
The whole arc on one page — useful for review.
| Era | Key Milestones | Why It Matters for Business |
|---|---|---|
| 1930s–1950s | Turing’s universal machine & Turing Test; first programs; Perceptron (1958); Dartmouth Conference (1956) | AI formally begins; the core ideas of computing, learning, and reasoning are established. |
| 1960s–1970s | Early reasoning programs; ELIZA (first chatbot); the first AI winter | Shows AI’s promise but exposes its limits; a lasting lesson in hype vs. reality. |
| 1980s–2000s | Expert systems; the rise of machine learning; the internet era (search, shopping, streaming) | AI delivers its first durable business value and becomes a profit engine. |
| 2010s | Image-recognition breakthrough (2012); Go mastered (2016); the transformer (2017); GPUs + big data | Deep learning transforms industries; AI becomes practical at business scale. |
| 2018–2022 | Large language models; the first mainstream chat assistant (2022) | Generative AI goes mainstream; businesses adopt it across departments. |
| 2023–2026 | Reasoning models; AI agents; rapid enterprise adoption | AI shifts from answering to acting — but results lag the hype, making judgment essential. |
Key Terms
Every important term from this chapter, in one place. Each entry gives a plain definition, the word used in a sentence, and a hint to help it stick. Review these before the quiz.
- Artificial Intelligence (AI)
- Definition: The broad goal of building machines that can perform tasks normally requiring human intelligence.
In a sentence: “The bank uses artificial intelligence to spot unusual transactions in real time.”
Hint: AI is the whole umbrella — everything else on this list fits underneath it. - Machine Learning (ML)
- Definition: A type of AI where systems learn patterns from data instead of following hand-written rules.
In a sentence: “Instead of programming every rule, the team let a machine-learning model find the patterns itself.”
Hint: The machine learns from examples, like a student with flashcards. - Deep Learning
- Definition: A type of machine learning that uses many-layered neural networks.
In a sentence: “Deep learning is what lets your phone recognize faces in photos.”
Hint: “Deep” = many layers stacked on top of each other. - Neural Network
- Definition: A computing system loosely inspired by the brain, built from layers of simple connected units.
In a sentence: “The neural network improved each time it saw more labeled images.”
Hint: Picture a brain made of math — lots of tiny connections working together. - Turing Test
- Definition: A test of machine intelligence: if a person cannot tell whether they are talking to a human or a machine, the machine passes.
In a sentence: “People still debate whether modern chatbots truly pass the Turing Test.”
Hint: “Can you tell who’s typing?” If not, the machine passes. - AI Winter
- Definition: A period when funding and interest in AI collapsed after the technology failed to meet its promises.
In a sentence: “After the hype of the 1980s came an AI winter that lasted years.”
Hint: Winter = a cold, frozen stretch where progress (and money) dries up. - Expert System
- Definition: An early AI approach that captured human expert knowledge as thousands of hand-written rules.
In a sentence: “The medical expert system suggested diagnoses by following its built-in rules.”
Hint: Like bottling one expert’s knowledge into a rulebook — powerful, but it can’t learn. - Perceptron
- Definition: An early (1958) trainable neural network — the seed of modern deep learning.
In a sentence: “The perceptron was decades ahead of its time.”
Hint: The tiny first ancestor of today’s giant neural networks. - Transformer
- Definition: The neural-network design, introduced in 2017, behind virtually every major AI system today.
In a sentence: “The transformer made it practical to train AI on enormous amounts of text.”
Hint: Not the movie robot — it “pays attention” to the most important words. - Large Language Model (LLM)
- Definition: An AI model trained on vast amounts of text to understand and generate human-like language.
In a sentence: “ChatGPT is powered by a large language model.”
Hint: “Large” = huge amounts of text; “language” = it works with words. - Generative AI
- Definition: AI that creates new content such as text, images, audio, or code.
In a sentence: “The marketing team used generative AI to draft three ad concepts in minutes.”
Hint: “Generative” → it generates something new. - Agentic AI
- Definition: AI that does not just answer questions but takes actions and completes multi-step tasks.
In a sentence: “An agentic AI can book the whole trip, not just suggest flights.”
Hint: An “agent” acts on your behalf — it does, not just says. - Reasoning Model
- Definition: An AI model that works through a problem step by step before answering, improving reliability on hard tasks.
In a sentence: “For complex math, a reasoning model outperforms a quick-answer chatbot.”
Hint: It “shows its work” like a careful student. - GPU (Graphics Processing Unit)
- Definition: A chip originally built for graphics that turned out to be ideal for training AI.
In a sentence: “Demand for GPUs soared as companies raced to train bigger AI models.”
Hint: Built for video games, now the engine room of AI. - ELIZA Effect
- Definition: People’s tendency to believe a machine understands them, even when it is following simple rules.
In a sentence: “Falling for a chatbot’s charm is the ELIZA Effect at work.”
Hint: Named after ELIZA, the 1966 chatbot that fooled people with simple tricks. - Narrow vs. General vs. Super AI
- Definition: Narrow AI does one task well (all of today’s AI); General AI would match humans across all tasks; Super AI would exceed humans. Only Narrow AI exists today.
In a sentence: “Despite the headlines, every system in use today is Narrow AI.”
Hint: Narrow = one lane; General = every lane; Super = beyond us. We’re still in the first lane.
Why History Matters
Across nearly a century, one rhythm repeats: soaring hype, hard setbacks, and then real, lasting progress. The expert systems failed, but their lessons fed the machine-learning era. The early neural networks disappointed, but the idea returned to power the deep-learning revolution. Each cycle left the field further ahead than the last.
Three points carry forward into the rest of this course. Today’s AI is a genuine source of competitive advantage — but only for organizations that deploy it wisely. The hype is real, and so are the limits; your value lies in telling them apart. And the lasting edge belongs to those who combine human judgment with AI’s analytical power, rather than surrendering one to the other.
Sources & further reading
These resources come from the world’s top-ranked AI universities and are good starting points if you want to go deeper:
- Stanford University (HAI) — What Is Artificial Intelligence? A clear, plain-language explainer from Stanford’s Institute for Human-Centered AI.
- Stanford University (HAI) — AI Index Report The most-cited annual, data-driven overview of AI progress and adoption.
- Carnegie Mellon University — History of AI at CMU A look at the field’s development from the university ranked #1 in AI.
- MIT OpenCourseWare — Artificial Intelligence MIT’s free foundational course; the first lecture covers the history of the field.