ONLINEAGENT_OPS 2026.Q3 HOME HUBSMODELSBUILDINGCOMPANIESFILMMAKINGSECURITYLIVING WITH ITSKILLSALL HUBS, ONE TERMINAL → ARTICLES CRAFT RECORD BLOG TOOLS SKILLS SEARCH /
HOME›ARTICLES›AGI MEANING
ARTICLES · EVERGREEN EXPLAINERBEGINNER

AGI Meaning: What Artificial General Intelligence Is (and Why Nobody Agrees)

One acronym, at least four competing definitions, and no agreed test. What each meaning claims, how AGI differs from today’s AI and from superintelligence, and how to read the headlines.

READ7 min
WORDS1,368
SECTIONS8
TYPEEXPLAINER
CHECKED25 AUG 26

AGI stands for artificial general intelligence: an AI that could match people across most thinking tasks, not just one narrow job. The 2023 “Levels of AGI” paper says “general” means a “wide range of non-physical tasks”.1 People disagree on where the line is and how to test for it.

(Not to be confused with AGI = adjusted gross income on tax forms.2)

"AGI" is used daily as if it named one thing. It doesn't. It is a contested label covering at least four distinct claims, and knowing which one a speaker means is the first step in evaluating sentences like "AGI is here" or "AGI is decades away." This piece maps the definitions, the history, the measurement problem, and the honest current state.

TL;DR — THE SHORT VERSION

"AGI" is a contested label, not one thing, so a claim that it is here or far away means little until you know which meaning is in use.

  • One word, four meanings. Human-level at nearly all thinking tasks, able to do most economically valuable work, able to pursue long goals reliably, or a ladder of levels rather than a finish line.
  • There is no agreed test. The Turing test, benchmarks and job-based tests have each failed or eroded, which is why headlines contradict each other.
  • Today's systems are broad but not reliable. The 2023 levels paper put the frontier models of its day at “Emerging AGI”; where today's systems sit on that scale is disputed.
  • AGI is not consciousness, and not a date. Capability and sentience are separate questions, and a published timeline is an argument, not a measurement.
  • Ask two questions. Which meaning, and what test would settle it? If the speaker cannot say, the claim is opinion.

Four competing definitions of AGI

The capability definition: a system matching or exceeding humans across virtually all cognitive tasks — the classic reading, tracing to Mark Gubrud's 1997 usage and popularized among computer scientists by Shane Legg and Ben Goertzel from 2001.34The economic definition: systems that can perform most economically valuable work — the framing used in OpenAI's charter.5 The autonomy definition: agents that can pursue long-horizon goals in the world with human-level reliability. The generality-gradient definition: AGI as a spectrum, not a threshold — formalized in the 2023 “Levels of AGI” paper by eight Google DeepMind researchers67 (emerging → competent → expert → exceptional → superhuman, crossed with narrow vs general). Level 4 was called “virtuoso” until the paper’s fifth version, in September 2025, renamed it “exceptional”.8

TAKEAWAY

Find out which meaning is in play. The same system can count as AGI under a loose definition and fall well short under the strict one.

AGI vs AI (narrow AI)

The plain point: almost all AI in use is “narrow”: built for one clearly scoped job, such as playing chess or predicting protein shapes, where it can beat any human. AGI would be general: good across a wide range of tasks, including learning new ones. The Levels of AGI paper makes the split explicit, listing chess and protein-folding systems as superhuman but narrow.1 Chatbots sit awkwardly in between: broad, but uneven. Narrow AI meaning has more.

AGI vs ASI (superintelligence)

The plain point: AGI means roughly human-level across the board; ASI, artificial superintelligence, means beyond every human at everything. In the Levels of AGI scale, the top level is defined as outperforming all humans, and its general version, ASI, is listed as not yet achieved.1 So ASI is a further claim on top of AGI, not another name for it. Superintelligence meaning has more.

TAKEAWAY

Narrow, general, super. Narrow AI does one job; AGI would do most thinking jobs about as well as people; ASI would do all of them better than anyone.

The measurement problem: there is no agreed test

Every proposed test has failed or eroded. The Turing test was arguably passed in spirit and dismissed as measuring deception, not intelligence. Benchmark suites saturate within months of release and leak into training data. Employment-based tests ("can it do most jobs?") depend on deployment and law as much as capability. The uncomfortable, factual position: there is no agreed instrument, which is why headlines can simultaneously announce AGI's arrival and its impossibility.

TAKEAWAY

Treat an "AGI has arrived" or "AGI is impossible" headline as an opinion about a test. Without an agreed instrument, neither can be settled by measurement.

Where things honestly stand

Frontier reasoning models and agent systems are broadly capable across text, code, vision, and increasingly action — clearing many pre-2020 informal AGI bars — while still failing at long-horizon reliability, genuine novelty under distribution shift, and consistent physical-world competence. On the levels framing, the paper's own authors placed the frontier language models of 2023 at Level 1, “Emerging AGI”, one step below “Competent AGI”.9 This site has found no agreed placement for 2026 systems on that scale, and there is sharp disagreement about the slope. Claims stronger than this, in either direction, outrun the evidence.

LEVELS OF AGI · A LADDER, NOT A FINISH LINE
The levels framing from 2023 treats AGI as a spectrum, each level crossed with narrow vs general.
EMERGINGWhere the 2023 paper placed the frontier language models of its day. Where 2026 systems sit is disputed.
COMPETENT
EXPERT
EXCEPTIONAL
SUPERHUMAN
Reasoning — redraws the levels named in this page’s first section, from Morris et al., “Levels of AGI”, arXiv:2311.02462, 2023 (Level 4, first called “virtuoso”, was renamed “exceptional” in the paper’s fifth version, September 2025), with the placement the paper itself gave for 2023 frontier language models.
TAKEAWAY

Be wary of confident claims in either direction. The evidence supports broad capability with real gaps in long-horizon reliability and novelty; anything stronger outruns it.

What AGI is not

Not consciousness — capability claims and sentience claims are independent. Not superintelligence — ASI is a further claim about exceeding all humans at everything. Not a product launch — no lab's announcement settles a definitional dispute. And not a date — every published timeline is an argument, not a measurement.

TAKEAWAY

Keep the questions apart. Whether a system is capable, whether it is conscious and when "AGI" arrives are separate claims, and a lab announcement settles none of them.

Questions people ask

Does AGI exist yet?

Not by any agreed test, because there is none. The Levels of AGI paper placed the frontier language models of 2023 at its lowest general level, “Emerging AGI”, and listed superintelligence as not yet achieved.1 This site has found no agreed placement for later systems.

Is ChatGPT AGI?

By the strictest meaning, no; by looser ones, partly. On the levels scale, ChatGPT-style models of 2023 counted as “Emerging AGI”: “Competent” at some tasks, “Emerging” at most.1 OpenAI’s charter uses the economic definition instead (see the four definitions above), which is a different bar again.

When will AGI arrive?

Nobody knows, and forecasts are arguments, not measurements. In a large survey of published AI researchers, released in 2024, the combined estimate gave a 10% chance of machines outperforming humans at every task by 2027 and 50% by 2047, thirteen years earlier than the same survey a year before.10 Some lab leaders are far more bullish: Anthropic’s chief executive, who says he dislikes the term AGI, wrote in October 2024 that what he calls “powerful AI” could come as early as 2026.11

Is AGI conscious?

That is a separate question: capability and consciousness are different claims, and a system could be very capable without any inner experience. Does AI understand anything? covers what researchers say about AI consciousness.

Is AGI the same as ASI?

No. ASI means beating all humans at everything, a further claim; see AGI vs ASI above.

Why the definition fight matters

Contracts, safety commitments, and even corporate governance provisions can hinge on the phrase. Regulation drafted against a word without an instrument becomes lobbying terrain. The practical takeaway for a reader of this site: whenever you meet the term, ask which of the four claims — and what test the speaker would accept as settling it. If they can't answer, you've learned what you needed.

TAKEAWAY

The word carries real stakes. Contracts and governance can hinge on it, so ask which meaning a speaker uses and what test they would accept as settling it.

Part of the Stay Human record. Related: how to ask AI well

Related on this site: Is AI Dangerous? · Types of AI, Explained · Turing test meaning · What Is the Singularity in AI?

Further reading on the definitional dispute itself: M. Mitchell, Debates on the nature of AGI, Science, 2024. Named, not opened: it returned 403 to an automated request on 16 Sep 2026, and on 5 Oct 2026 science.org showed a bot challenge, which this site does not bypass. Nothing on this page rests on it.

SOURCES
  1. 1Morris et al., “Levels of AGI”, arXiv:2311.02462, 2023, read at source 5 Oct 2026the “General” column covers a “wide range of non-physical tasks, including metacognitive tasks like learning new skills”; the “Narrow” column a “clearly scoped task or set of tasks”. ↑aNarrow: “clearly scoped task or set of tasks”; the narrow systems listed at the superhuman level are “AlphaFold”, “AlphaZero” and “StockFish”. ↑bLevel 5, “Superhuman”: “outperforms 100% of humans”; the general entry, “Artificial Superintelligence (ASI)”, is marked “not yet achieved”. ↑c“Overall, current frontier language models would therefore be considered a Level 1 General AI (‘Emerging AGI’) until the performance level increases for a broader set of tasks”. ↑d“as of this writing in September 2023, frontier language models (e.g., ChatGPT, Bard, Llama2, etc.) exhibit ‘Competent’ performance levels for some tasks (e.g., short essay writing, simple coding), but are still at ‘Emerging’ performance levels for most tasks (e.g., mathematical abilities, tasks involving factuality).” ↑e
  2. 2Internal Revenue Service, Definition of adjusted gross income, read at source 5 Oct 2026“Your adjusted gross income (AGI) is your total (gross) income from all sources minus certain adjustments listed on Schedule 1 of Form 1040.” ↑
  3. 3M. Gubrud, Nanotechnology and International Security, 1997 — the first documented use of “artificial general intelligence”. Archived copy; link checked 16 Sep 2026.↑
  4. 4Morris et al. (Google DeepMind), Levels of AGI for Operationalizing Progress on the Path to AGI, arXiv:2311.02462v5, 24 Sep 2025, read at source 5 Oct 2026“Legg (Legg, 2008) and Goertzel (Goertzel, 2014) popularized the term AGI among computer scientists in 2001 (Legg, 2022)” ↑
  5. 5OpenAI, OpenAI Charter, 2018, the source of the economic definition. Link checked 16 Sep 2026.↑
  6. 6Morris, Sohl-Dickstein, Fiedel, Warkentin, Dafoe, Faust, Farabet & Legg, “Levels of AGI”, arXiv:2311.02462, 2023. Link checked 16 Sep 2026.↑
  7. 7Morris et al., Levels of AGI for Operationalizing Progress on the Path to AGI, arXiv:2311.02462v5, 24 Sep 2025, read at source 5 Oct 2026the author list gives each of the eight authors the affiliation “Google DeepMind”, starting “Meredith Ringel Morris Affiliation: Google DeepMind, Seattle, WA, USA” ↑
  8. 8Morris et al., Levels of AGI for Operationalizing Progress on the Path to AGI, arXiv:2311.02462, v5 submitted 24 Sep 2025, read at source 5 Oct 2026“version 5 - We updated the nomenclature of Level 4 to be ‘Exceptional’ instead of ‘Virtuoso’”; the paper’s footnote: “While Level 4 was originally called ‘Virtuoso AGI,’ we now use the term ‘Exceptional AGI,’” ↑
  9. 9Morris et al., “Levels of AGI”, arXiv:2311.02462 (v5), read at source 17 Sep 2026“Overall, current frontier language models would therefore be considered a Level 1 General AI (“Emerging AGI”) until the performance level increases for a broader set of tasks (at which point the Level 2 General AI, “Competent AGI,” criteria would be met).” An earlier version of this page said “credible observers” place today’s systems around “competent-general”; no source was given, and none was found. ↑
  10. 10Grace, Stewart, Sandkühler, Thomas, Weinstein-Raun, Brauner & Korzekwa, Thousands of AI Authors on the Future of AI, Jan 2024 (revised Oct 2025), read at source 5 Oct 2026“If science continues undisrupted, the chance of unaided machines outperforming humans in every possible task was estimated at 10% by 2027, and 50% by 2047.” and “The latter estimate is 13 years earlier than that reached in a similar survey we conducted only one year earlier.” ↑
  11. 11Dario Amodei, Machines of Loving Grace, Oct 2024, read at source 5 Oct 2026“I think it could come as early as 2026, though there are also ways it could take much longer.” and “I dislike the term AGI”. ↑

Free printable AI cheat sheets, and AI news every week.

Sixteen one-page AI guides and checklists in one free PDF, no sign-up. And if you want to keep up, the newsletter covers what changed in AI each week, in plain English, with the sources.

SUBSCRIBE FREE ↗ GET THE CHEAT SHEET PACK