AI Search 101

What is artificial intelligence?

The basics

Abbreviation	AI
Named	1956, at the Dartmouth workshop
Main modern approach	Machine learning, particularly deep learning
Includes	Generative AI, computer vision, language processing, robotics, planning
Key distinction	Strong and narrow AI
Classic test	The Turing test

Why is artificial intelligence so hard to define?

Because the definition moves. Tasks are called artificial intelligence until they work reliably, at which point they are called software.

Optical character recognition, chess, spam filtering, speech transcription and route planning were all landmark AI problems. None is described as artificial intelligence now. They became ordinary features, and the label moved on to whatever remained difficult. This pattern is old enough to have a name, the AI effect, and it explains why the field can appear to make no progress while progressing steadily.

The practical consequence for a business is that the word carries almost no information on its own. What matters is which technique, applied to which task, with what error rate.

How did the field get here?

Early work, from the 1950s into the 1980s, tried to encode intelligence as explicit rules. Experts wrote down what they knew and systems applied it. This worked in narrow domains and collapsed under real-world variety, because the number of rules required grows faster than anyone can write them.

The shift that produced everything current was giving up on writing the rules. Instead of specifying how to recognise a cat, show the system enough labelled examples that it derives the pattern itself. That is machine learning, and from the 2010s it was combined with much larger neural networks, far more data and far more computing power. Large language models are the current expression of that approach.

What are the main branches?

Machine learning covers systems that learn patterns from data. Deep learning is machine learning using many-layered neural networks, and is behind most recent progress. Natural language processing deals with human language. Computer vision deals with images and video. Robotics deals with acting in the physical world. Planning and search deal with finding sequences of actions toward a goal.

These overlap heavily. A modern assistant draws on several at once, which is another reason the single label is unhelpful.

Is any of this actually intelligent?

A genuine and unresolved dispute, and worth stating carefully rather than settling.

Current systems perform tasks that plainly required intelligence when a person did them. They also fail in ways no person would, have no understanding of what they are doing in any sense involving experience, and cannot transfer competence between domains the way a person can.

Whether that adds up to intelligence depends on whether intelligence is defined by capability or by the nature of the underlying process. Both positions are held by serious people. What can be said without controversy is that these are narrow systems: extremely capable within a domain, and nothing outside it.

How artificial intelligence is used in marketing

Production. Generating and adapting text, images and video, and translating between languages.

Analysis. Segmentation, propensity and churn modelling, attribution, and pulling structure out of open-text feedback. Much of this predates the current wave and is simply called analytics.

Distribution. Bidding and targeting inside advertising platforms, recommendation systems, and now assistants that answer buyers' questions about categories and vendors directly.

How this impacts your business

A language model does not look a company up in a register. It composes a description from everything it read during training and whatever it retrieves at the moment of the question. Trade press, directories, forums, competitor comparison pages, and somewhere in that mix the company's own website too.

So the description a buyer sees is assembled, not published. No company controls it, but every company supplies a share of the material it is built from.

The traffic impact is still small, roughly one percent of sessions for most companies. The influence is not small, because the reader is often a buyer forming a first impression of what a company does and who it serves, before any page gets a click.

Checking takes ten seconds. Type the company name into ChatGPT and read the answer.

Frequently asked questions

What is the difference between AI and machine learning?
Machine learning is one approach within artificial intelligence, specifically learning patterns from data. Artificial intelligence is the broader field and includes rule-based methods that involve no learning at all.

When was artificial intelligence invented?
The term was coined for a 1956 workshop at Dartmouth College, though the underlying questions are older. The field has passed through several cycles of enthusiasm and disappointment since.

Is artificial intelligence going to replace jobs?
It is changing which tasks within jobs are done by people. Roles heavy in routine information work are affected first. The historical pattern is redistribution rather than wholesale elimination, though the transition is rarely comfortable for those in it.

What is artificial general intelligence?
A hypothetical system with broad human-level capability across domains. Nothing existing approaches it, and forecasts of its arrival vary by decades. See the technological singularity.

Does a business need an AI strategy?
It needs decisions about specific tasks, tools and data handling. A strategy document about artificial intelligence in general tends to produce less than a clear answer about three concrete use cases.

Related definitions

• What is machine learning?
• What is deep learning?
• Strong and narrow AI
• What is generative AI?
• AI Search 101

---

Written by Lari Numminen, Generate More. Also available in Finnish. Updated 31 August 2026.