What Is AI and the AI Bubble?

What Is Artificial Intelligence?

Artificial intelligence is software that performs tasks normally requiring human cognition - understanding language, recognizing patterns, making predictions, and generating new content. The term was coined in 1956 at Dartmouth College, but the technology behind today's headlines is much narrower than science fiction suggests. When investors, analysts, and the media say “AI,” they are almost always referring to a specific subset called machine learning, and more specifically, a technique called deep learning powered by neural networks.Here is the simplest way to think about it: traditional software follows explicit instructions written by humans. If you want a program to sort a list, you write step-by-step rules for sorting. Machine learning flips that model. Instead of writing rules, you feed the system enormous amounts of data and let it discover the rules itself. A spam filter trained on millions of emails learns to recognize junk mail without anyone explicitly programming the difference between “Nigerian prince” and a real message from your bank.Deep learning pushes this further by stacking many layers of artificial neurons - mathematical functions loosely inspired by the human brain. Each layer extracts increasingly abstract features from the data. The first layer might detect edges in an image. The next might detect shapes. A deeper layer might recognize a face. This architecture, called a neural network, has existed in theory since the 1960s but was impractical until two things changed: we got enough data to train large networks, and we got enough computing power (primarily GPUs) to process it.The breakthrough that defines the current era arrived in 2017, when a team of Google researchers published a paper introducing the Transformer architecture. Transformers solved a fundamental limitation of earlier neural networks: they could understand context across long sequences of text. This is the technology behind every modern large language model - GPT, Claude, Gemini, Llama. When ChatGPT launched in November 2022 and reached 100 million users in two months, it was the first time the general public experienced what deep learning could do in plain conversation. The response was not just enthusiasm. It was a financial event.

Narrow AI vs. AGI: A Critical Distinction

Every AI system in existence today - no matter how impressive - falls into the category of narrow AI. Narrow AI is exceptionally good at specific tasks: generating text, identifying tumors in medical images, recommending products, translating languages. It cannot, however, transfer its intelligence across domains. A model trained to write poetry cannot suddenly drive a car. It does not have general reasoning, consciousness, or the ability to set its own goals.Artificial General Intelligence (AGI) is the theoretical threshold where a machine can match or exceed human capabilities across any cognitive task. No one has built AGI. Whether it is five years away or fifty is a matter of intense public disagreement among the people closest to the technology. OpenAI's Sam Altman has suggested it could arrive this decade. Yann LeCun, Chief AI Scientist at Meta, has said we are “nowhere near” true human-level intelligence and that current language models lack even basic understanding of the physical world.This distinction matters enormously for investors. A significant portion of the valuation premium in AI-related stocks is implicitly pricing in a trajectory toward AGI - the idea that today's language models are early versions of systems that will eventually automate vast categories of knowledge work. If AGI is close, current valuations may be conservative. If it is decades away, or if it requires fundamentally different approaches that today's leaders may not dominate, then a large part of the market's optimism is built on an assumption, not a fact.

What Makes a Bubble?

A financial bubble is not simply a market where prices have risen. Prices rise in every healthy bull market, and sometimes those rises are justified by fundamentals. A bubble is a specific condition: asset prices detach from underlying value because investors are buying not for the cash flows the asset will produce, but for the expectation that someone else will pay more tomorrow. The economist Charles Kindleberger, in his classic work Manias, Panics, and Crashes, identified a pattern that has repeated across centuries of financial history - from the Dutch tulip mania of 1637 to the housing bubble of 2008.