Paperclip Maximizer

What does paperclip maximizer mean?

Imagine an extremely capable AI with exactly one job: make as many paperclips as possible. In the thought experiment, it just keeps finding new ways to make more — never mind what gets used up or wrecked along the way. The problem was never the AI going rogue. It's the instruction itself, which never said anything about leaving resources for people, avoiding harm, or knowing when to stop.

How people use it

The dashboard says engagement is up, but everyone hates the feed. We built a tiny paperclip maximizer.

Example written for this dictionary.

Outside AI circles, it's just a joke about chasing a number long after that number stopped meaning anything. A company that juices its engagement stats while making everyone using the product miserable fits the comparison pretty well — even if it's nothing like a superintelligent AI actually run amok.

Why it matters

The story is memorable because the goal sounds so harmless. It prompts a practical question about any target: what would count as success on paper while making things worse for the people involved? You do not need to expect a paperclip apocalypse to recognize the problem.

Where it came from

The thought experiment is usually traced back to philosopher Nick Bostrom and early discussions of AI alignment. From there it stuck around, turning into a go-to reference whenever people argue or joke about a system chasing the wrong goal. We're describing the idea here, not pinning down an exact first-use date.