AGI
/ˌeɪ.dʒiːˈaɪ/
UncommonArtificial General Intelligence. A hypothetical AI that can do any cognitive task a human can.
Increasingly debated as a near-term claim. Some labs (OpenAI, DeepMind) have redefined or hedged their definitions of AGI in ways that critics see as moving goalposts. The term is now frequently used sarcastically when a lab makes vague capability claims.
| # | Example | What it means | Source |
|---|---|---|---|
| 01 | Every 6 months someone says AGI is 2 years away. It has been 2 years away for a decade. | Mocking the repeated pattern of near-term AGI predictions that never land. | x, 2024 |
| 02 | They quietly redefined AGI so they could claim they already built it. Classic. | Accusing a company of shifting the definition of AGI to fit their current system. | reddit, 2024 |
| 03 | I honestly do not know if we are 5 years or 50 years from AGI and anyone who says they do is lying. | Expressing honest uncertainty about AGI timelines. | x, 2024 |
| 04 | If we do build AGI without solving alignment first, we are cooked. | Warning that AGI without safety guarantees is dangerous. | reddit, 2023 |
Use it when
- Talking about the theoretical milestone of AI matching human-level general intelligence.
- Discussing whether current AI systems count as AGI or are approaching it.
- Referring to lab claims about their own systems reaching AGI.
Do not use it when
- You just mean a really capable AI model or chatbot.
- You mean AI that is smarter than a human in a narrow domain like chess or Go.
Variations and related forms
AGI vs superintelligence
AGI
AGI matches human-level general intelligence across all domains.
superintelligence
Superintelligence goes beyond AGI to far exceed human capabilities, potentially in ways humans cannot comprehend.
Origin and context
The term AGI was coined to distinguish general AI from narrow AI. It became central to AI safety discourse in the 2000s through researchers like Ben Goertzel and gained mass awareness through OpenAI's stated mission to build AGI safely.
- First seen
- early-2000s
- Confidence
- medium
Cite this page
Use this stable URL to reference the entry.
https://slang.fyi/agi