Artificial intelligence

(systems that learn statistical patterns from data to perform tasks associated with human intelligence)

What is artificial intelligence?

Artificial intelligence (AI) is the field concerned with building systems that perform tasks normally associated with human intelligence — recognizing patterns, making decisions, understanding language — usually by learning from examples rather than following a fixed list of rules written in advance.

Rule-based systems versus learning from data

An early approach to AI was rule-based: a programmer writes out explicit rules by hand ("if the word 'urgent' appears three times, flag this email"). This works for simple, well-understood problems, but breaks down quickly for anything complex — no one can hand-write every rule that distinguishes a photo of a cat from a photo of a dog. Modern AI instead usually learns its own rules directly from a large set of labeled examples.

A concrete example: a spam filter

Give a program thousands of emails, each already labeled "spam" or "not spam." Rather than being told explicit rules, the program looks for patterns in the labeled examples — certain words, phrases, or combinations that show up disproportionately often in the spam-labeled emails. Once trained this way, it can look at a brand-new, previously unseen email and estimate how likely it is to be spam, based on how closely its patterns match what it learned from the labeled examples.

Classification: drawing a boundary between categories

Many AI tasks boil down to classification — sorting an input into one of several categories. If you plotted every training email as a point, based on how often certain key words appear, spam and non-spam emails would tend to cluster in different regions. The learning process amounts to finding a boundary that best separates those regions, so that a new email can be classified by simply checking which side of the boundary it falls on.

Why AI matters — and what it actually is

AI already powers real, established tools used daily: search engines ranking results, streaming services recommending content, and photo apps recognizing faces. It's worth being precise about what these systems are actually doing: they are finding statistical patterns in the specific data they were trained on, and applying those patterns to new inputs — a genuinely powerful and useful technique, but a different thing from claims about human-like understanding or general reasoning, which remain open questions actively debated by researchers rather than settled facts.