Clicked Gallery
What is AGI (artificial general intelligence)?
Highlighted from a real engineering doc. Explained by Clicked.
Used in a sentence
Engineering Notes · AI Systems
Investors pressed the executives on when the company expected to reach AGI, and no one would commit to a date.
The reader highlighted one word in the docs. Clicked explained the technical term “AGI” in simple terms:
Explained in three depths
●○○
Overview
AGI, short for artificial general intelligence, is a hypothetical AI that could handle any intellectual task a person can: learn a new field, plan, write, reason, and switch between all of them. No such system exists. There is also no agreed test for one, which is why experts can look at the same AI and disagree about how close it is.
●●○
Detail
AGI is the name for an AI that could do everything a person can do with their mind: any subject, any task, including ones it was never trained for. It is a destination, not a shipped product. Everything deployed so far is narrower than that. The word exists because "AI" alone stopped being useful. A spam filter is AI, a route planner is AI, and neither can do the other's job, so the field needed a word for breadth itself. That is the claim AGI makes: one system, every desk. The catch is that nobody agrees where the line sits. One camp defines it by skill, whether it matches human experts across subjects. Another defines it by output, whether it could do most work people are paid to do. There is no agreed exam, so one researcher says AGI is decades away, another says it is nearly here, and both point at the same systems. Two things it does not mean. AGI is not superintelligence, the word for systems beyond the best humans at essentially everything. And general does not mean conscious. It is a claim about what a system can do, not about whether anything is going on inside.
●●●
Analogy
AGI would be a cook, and everything built so far is kitchen equipment. A bread machine has one move. A modern chatbot is more like a food processor with a shelf of attachments: it slices, kneads, and juices, and it can feel like it does anything. The attachments still decide what it can do, and it does not notice a burning pan. A cook can walk into a strange kitchen, read the situation, and manage a dish nobody wrote down. The argument in AI is how far the attachments are from the cook. So far, every machine in the kitchen is still equipment.
AI explanations may contain errors · Not professional advice
Same facts, different vibe — Slang mode 😎
●○○
Overview
AGI is the everything-AI: one system that could take on whatever a person can, not one brilliant trick per machine. Nobody has built one. Better, nobody has agreed on what would count as building one, so the industry argues about how close it is with a straight face. Half of AI news is people racing toward a finish line that has never been drawn. 😎
●●○
Detail
AGI is the everything-AI that does not exist yet, and the word is carrying serious money in the meantime. Roadmaps promise it, funding rounds price it, and there are contracts where real obligations turn on whether the label applies. Now the awkward part: the label has no agreed test. Ask ten researchers what would count and you get ten different exams, from matching experts in every field to doing most office work unsupervised. So when someone announces AGI is two years away, there is no exam to grade that claim against, and no way for you to check it either. Keep that in mind whenever a headline uses the word. There is also an old habit at work underneath, and it has a name, the AI effect: once software masters a skill, people quietly stop counting that skill as intelligence. Machine translation is the classic case. AGI inherits the habit in its purest form. The word names whatever machines still cannot do, and that list gets rewritten after every demo. 😎
●●●
Analogy
Narrow AI is a sprinter: one event, world record, done. AGI would be a decathlete, at least solid at every event, including the one the committee invented this morning. The complication is the judging. The federation never published what counts as passing, so every athlete competes against a rulebook that does not exist. And each time an event gets cleared, someone in the booth explains that this event was never really the point. Until the rulebook exists, the argument is the sport. 😎
AI explanations may contain errors · Not professional advice
Formal definition — The same term, explained the usual way
Artificial general intelligence denotes a hypothetical class of artificial systems able to match or exceed human cognitive performance across the full range of intellectual tasks, including learning and transfer to unfamiliar domains, rather than within a delimited task distribution. The term lacks a single operational definition; proposed criteria range from parity with human professionals across fields to autonomous performance of most economically valuable work. It is distinguished from narrow artificial intelligence, which is confined to specified tasks, and from superintelligence, which denotes performance beyond the best human level in essentially all domains.
Want Clicked to explain terms like “AGI” directly in your browser — including on PDFs?
Add to Chrome — Free50 free Explanations · No credit card required
More from the gallery
What is a large language model (LLM)?
A program that predicts what text comes next, then taught separately to answer rather than ramble.
Machine learning vs AI: what's the difference?
AI is the goal. Machine learning is the method that took over. The words are not interchangeable.
What is recursive self-improvement?
An AI helps build the next AI, which builds the one after that, better each round.
What are AI Benchmarks?
The scores behind every AI launch, what they measure, and why the leader can still disappoint.
What is training (in AI)?
The process that builds a model by example. Chatting with it afterwards changes nothing.
What is Inference (in AI)?
The model doing its actual job, and why the cheap half ends up costing the most.