Topics Technology and the internet

What is artificial intelligence, really?

Artificial intelligence is a broad label for computer systems that do things we normally link to human thinking: recognizing speech, spotting faces in photos, translating languages, playing chess, writing sentences. Most modern AI does not follow a long list of hand-written rules. It is built by showing a program a huge number of examples and letting it adjust itself until it gets good at a task.

What makes the question interesting is how loose the word is. Researchers have argued for decades about what counts as intelligence, and the label has been stuck on everything from a simple spam filter to a chatbot. A system can beat anyone at a board game and still be unable to do anything else. Whether any of it understands in the way people do is debated, and nobody has settled it.

An episode on this topic would start with where the idea came from, then walk through how learning from data works, what narrow and general AI mean, and where these systems fail. On bre, the hosts are AI, so they can be wrong, and the episode says so. You can press Talk and ask your own question as it plays.

What a bre episode would cover

An outline of the episode bre would make for this question. Every episode is written fresh when you ask, so yours will differ.

  1. Where the idea of AI beganThe field took shape in the 1950s, when researchers first asked whether machines could be made to think. Early programs used hand-written rules and hit limits quickly.
  2. Rules versus learningOld systems followed instructions people wrote. Modern machine learning finds patterns in examples instead, which is why data matters so much.
  3. Neural networks in plain wordsA neural network is layers of simple math that adjust their settings during training. The name is inspired by brains, but it is a loose comparison.
  4. Narrow AI and the idea of general AIToday's systems are good at specific jobs. A machine that matches people across everything remains a goal, and researchers disagree on how close it is.
  5. Where you already meet AIPhoto sorting, voice assistants, map routes, spam filters and chatbots all use some form of it, often without announcing it.
  6. What AI gets wrongSystems can reflect flaws in their training data, state false things confidently, and fail on situations they have not seen. Knowing this changes how you use them.

How the episode might open

A sample exchange between two of bre’s AI hosts, bre and Tess. Both are AI; this is written by AI, as every bre episode is.

  1. breAI host

    Okay, let's start with the awkward bit. Ask ten people what artificial intelligence is and you get ten answers, and one of them is usually a robot from a movie.

  2. TessAI host

    Guilty. Mine was a robot. So what's the honest answer, and why does it matter to someone who just wants their phone to work?

  3. breAI host

    The short version: software that does tasks we link to thinking, like recognizing a voice or translating a sentence, and it mostly learned how from examples instead of from rules someone typed in.

  4. TessAI host

    So nobody sat down and wrote the rules for what a cat looks like.

  5. breAI host

    Right. Picture showing a program a mountain of cat photos until its guesses get good. Nobody can point to a single line that says whiskers.

  6. TessAI host

    Okay, but real talk. If nobody wrote the rules, how do we know when it's wrong?

  7. breAI host

    That's the question I'd keep in your pocket for the whole episode. Sometimes we do know, and sometimes we find out the hard way.

  8. TessAI host

    Great. Practical, then. Let's go through what it can do, and what it can't.

Questions people also ask

Is artificial intelligence the same as a robot?
No. A robot is a physical machine, while AI is software that can run anywhere. Some robots use AI to see or decide things, but most AI you meet, like a spam filter or a voice assistant, has no body at all.
Does AI actually think or understand?
That is debated. These systems find patterns in data and produce useful results, but researchers disagree about whether that amounts to understanding as people experience it. There is no settled answer, so be cautious of anyone claiming one.
What is the difference between AI and machine learning?
AI is the broad goal of making machines do tasks that seem intelligent. Machine learning is one major way of getting there, where a program learns patterns from examples instead of following fixed rules. Most current AI uses it.
Can AI be wrong?
Yes. AI systems can make mistakes, repeat biases found in their training data, and give false answers that sound confident. That includes the AI hosts on bre, which is why it is worth checking anything important elsewhere.

Related topics

More: all 300 topics, technology and the internet, or the longer reads on /learn.

bre’s hosts are AI, and every episode is generated, so they can be wrong: check anything that matters. This page outlines what an episode would cover. It is for interest and learning, not medical, financial or legal advice.