What is AI?

What is AI?

Artificial intelligence (AI) involves the study and creation of computer systems that simulate human intelligence. AI is an extensive field, more than half a century old, covering everything from logic-based tasks to complex learning capabilities involving neural networks and deep learning. The term was coined in 1956 during the Summer Research Project on Artificial Intelligence at Dartmouth University. This was birth of modern AI research, in which researchers aimed to create systems that mimic human reasoning. However, this approach was rooted in the development of expert systems designed to solve specialized problems. Early AI thus looked a little different.

Recent breakthroughs in AI have been driven by two things: the advancements in computing power and the availability of massive datasets. Modern AI predominantly uses machine learning (ML), a subfield that enables algorithms to improve automatically by processing massive datasets instead of relying on rafts of programmers writing code.

There are several branches within machine learning: supervised learning (where algorithms learn from labeled examples), unsupervised learning (where systems identify patterns within unlabeled data), and reinforcement learning (which relies on feedback loops to guide behavior). Neural networks (modeled on the structure of the human brain), currently at the forefront of AI research, are capable of handling complex tasks such as image and speech recognition. AI's applications span nearly every industry today—from healthcare diagnostics and predictive financial analytics, to autonomous vehicles and natural language processing (NLP) used in type- and voice chat clients, for example.

It's tempting to think of AI reductively— as just a digital assistant that is now being built into almost every piece of technology we use. If it seems to be showing up everywhere, it's because AI is very good at quickly learning and taking over tasks that humans used to do, such as recognizing faces, recommending products, or answering customer service questions. However, the capabilities of AI continue to grow exponentially, surfacing a host of important political, cultural, ethical, economic, and other impacts centered around this disruptive new technology. One thing is clear—as machines increasingly handle tasks traditionally performed by humans, AI brings with it a raft of transformations that must be reckoned with at a societal level. Just how powerful is AI?

Types of AI

Before diving into categories, it's important to clear up what AI isn't. Today's AI systems are not all-powerful, sentient beings endowed with humanlike understanding or consciousness. They don't have desires, beliefs, or self-awareness. They simply follow statistical patterns learned from data. Unlike the omniscient AI villains or benevolent supercomputers of Hollywood, real-world AI is narrow in scope—it excels at specific tasks for which it's trained, but it lacks common-sense reasoning and cannot generalize knowledge in the open-ended way a human can. AI can't feel emotions, form intentions, or truly “understand” context beyond what its data allows.

AI is typically divided into three categories, ranging from current- to estimated future capabilities:

By understanding what AI can do (highly accurate pattern recognition, rapid data processing, narrow automation) and what it can't do (generalize like a human, possess consciousness or intent), we can better appreciate both its transformative potential and its current limitations.

For more information, consult this curated list of 'must-watch' TED Talks on AI.

Generative AI