Loading
What changed between revision 1 and revision 3, compared over the article's markdown source with its frontmatter — so a change to status, evidence or reversibility shows up as the edit it is.
Older · revision 1
Import· initial corpus import, not an edit
Initial import of content/artificial-general-intelligence.md — the filesystem corpus, unchanged. Not an edit.
14.3 kB
Revision 1
Revision 3
lines 7–13 → 7–136 unchanged lines not shown
categories: ["foundations", "minds"]categories: ["foundations", "minds"]tags: ["artificial intelligence", "agi", "forecasting", "benchmarks", "machine learning", "risk"]tags: ["artificial intelligence", "agi", "forecasting", "benchmarks", "machine learning", "risk"]summary: "A hypothetical artificial system matching or exceeding competent human performance across essentially all cognitive tasks rather than in narrow domains."summary: "A hypothetical artificial system matching or exceeding competent human performance across essentially all cognitive tasks rather than in narrow domains."updated: "2026-07-27"updated: "2026-10-29"issues: ["The 2024 ARC-AGI result and its harder 2025 successor set are described without a source.", "The EU AI Act compute thresholds are described without citing the instrument."]issues: ["The 2024 ARC-AGI result and its harder 2025 successor set are described without a source.", "The EU AI Act compute thresholds are described without citing the instrument."]------ lines 30–36 → 30–3616 unchanged lines not shown
```keyfacts```keyfacts[[ { "value": "1997", "label": "First recorded use of the phrase", "note": "Mark Gubrud, in a paper on nanotechnology and international security" }, { "value": "1997", "label": "First recorded use of the phrase", "note": "Mark Gubrud, in a paper on nanotechnology and international security" }, { "value": "~6 months", "label": "Doubling time of frontier training compute", "note": "estimated over 2010–2022 by Sevilla et al." }, { "value": "~6 months", "label": "Doubling time of frontier training compute", "note": "estimated over 2010-2022 by Sevilla et al." }, { "value": "6", "label": "Performance levels in DeepMind's AGI framework", "note": "'no AI' through 'superhuman', crossed with an axis of generality" } { "value": "6", "label": "Performance levels in DeepMind's AGI framework", "note": "'no AI' through 'superhuman', crossed with an axis of generality" }]]``````lines 39–47 → 39–472 unchanged lines not shown
There is no accepted definition. Legg and Hutter, surveying the problem, assembled more than seventy published definitions of intelligence and extracted a common core: intelligence measures an agent's ability to achieve goals in a wide range of environments.[^legg2007] That formulation is precise enough to be formalised and too abstract to adjudicate any actual system.There is no accepted definition. Legg and Hutter, surveying the problem, assembled more than seventy published definitions of intelligence and extracted a common core: intelligence measures an agent's ability to achieve goals in a wide range of environments.[^legg2007] That formulation is precise enough to be formalised and too abstract to adjudicate any actual system. Working definitions split into three families. *Task-based* definitions specify jobs a system must do — the Turing test is the ancestor, and modern versions substitute economically valuable work or a benchmark suite. *Capability-based* definitions specify cognitive functions such as transfer learning, long-horizon planning, or acquiring new skills without retraining. *Autonomy-based* definitions require operation over extended periods without human scaffolding. [[nick-bostrom]] sidesteps the threshold in *Superintelligence*, defining superintelligence as intellect greatly exceeding human performance in virtually all domains and treating human-level capability as a waypoint.Working definitions split into three families. *Task-based* definitions specify jobs a system must do. The Turing test is the ancestor; modern versions substitute economically valuable work or a benchmark suite. *Capability-based* definitions specify cognitive functions such as transfer learning, long-horizon planning, or acquiring new skills without retraining. *Autonomy-based* definitions require operation over extended periods without human scaffolding. [[nick-bostrom]] sidesteps the threshold in *Superintelligence*, defining superintelligence as intellect greatly exceeding human performance in virtually all domains and treating human-level capability as a waypoint. A framework from Google DeepMind researchers replaces the binary with a grid, scoring systems on performance (from "emerging" through "competent", "expert", "virtuoso" and "superhuman") crossed with generality.[^morris2024] On that scheme, mid-2020s systems are plausibly "emerging AGI" and clearly superhuman narrow AI in several domains — more informative than either "AGI is here" or "AGI is not here".A framework from [Google DeepMind](https://aiwiki.ai/wiki/google_deepmind) researchers replaces the binary with a grid, scoring systems on performance (from "emerging" through "competent", "expert", "virtuoso" and "superhuman") crossed with generality.[^morris2024] On that scheme, mid-2020s systems are plausibly "emerging AGI" and clearly superhuman narrow AI in several domains, a description more informative than either "AGI is here" or "AGI is not here". > [!note] Terminology> [!note] Terminology> "AGI" is used inconsistently even within single organisations: human-level competence across tasks, full automation of remote work, or the ability to do novel science. These are different claims with different timelines, and arguments equivocate between them.> "AGI" is used inconsistently even within single organisations: human-level competence across tasks, full automation of remote work, or the ability to do novel science. These are different claims with different timelines, and arguments equivocate between them.lines 50–64 → 50–642 unchanged lines not shown
Alan Turing's 1950 paper posed machine thinking in behavioural terms and proposed the imitation game to sidestep definitional argument.[^turing1950] The 1956 Dartmouth proposal assumed generality as the default goal, expecting machines to use language, form abstractions and improve themselves within a summer.Alan Turing's 1950 paper posed machine thinking in behavioural terms and proposed the imitation game to sidestep definitional argument.[^turing1950] The 1956 Dartmouth proposal assumed generality as the default goal, expecting machines to use language, form abstractions and improve themselves within a summer. The narrower ambitions of expert systems and, later, statistical machine learning displaced that goal for decades. The phrase "artificial general intelligence" appears in a 1997 paper by Mark Gubrud on the security implications of advanced technology, and was independently adopted in the early 2000s by Shane Legg, Ben Goertzel and Peter Voss, becoming established with a 2007 edited volume of that title. The phrase then distinguished a small community from a mainstream that regarded the goal as disreputable; several of the largest laboratories now state it as their objective.The narrower ambitions of expert systems and, later, statistical machine learning displaced that goal for decades. The phrase "artificial general intelligence" appears in a 1997 paper by Mark Gubrud on the security implications of advanced technology, and was independently adopted in the early 2000s by [Shane Legg](https://aiwiki.ai/wiki/shane_legg), Ben Goertzel and Peter Voss, becoming established with a 2007 edited volume of that title. The phrase then distinguished a small community from a mainstream that regarded the goal as disreputable; several of the largest laboratories now state it as their objective. ## Measuring progress## Measuring progress Benchmark performance is the field's main evidence, and it has an unusual failure mode: benchmarks saturate faster than they can be built. Knowledge tests, grade-school mathematics sets and function-level coding tests that were discriminating in the early 2020s reached ceiling within a few years. Their harder successors — graduate-level science, research mathematics, adversarially collected expert questions, repository-scale software tasks — have compressed in turn.Benchmark performance is the field's main evidence, and it has an unusual failure mode: benchmarks saturate faster than they can be built. Knowledge tests, grade-school mathematics sets and function-level coding tests that were discriminating in the early 2020s reached ceiling within a few years. Their harder successors, covering graduate-level science, research mathematics, adversarially collected expert questions and repository-scale software tasks, have compressed in turn. Saturation is weaker evidence than it appears. Contamination is hard to exclude when training corpora are web-scale and benchmarks are published; benchmarks measure what is easy to score, biasing them toward short checkable answers; and Goodhart's law applies with force when the benchmark is also the optimisation target for the labs reporting the scores.Saturation is weaker evidence than it appears. Contamination is hard to exclude when training corpora are web-scale and benchmarks are published; benchmarks measure what is easy to score, biasing them toward short checkable answers; and Goodhart's law applies with force when the benchmark is also the optimisation target for the labs reporting the scores. The most-cited contamination-resistant test is François Chollet's Abstraction and Reasoning Corpus, built around tasks requiring a novel rule to be inferred from a handful of examples.[^chollet2019] It resisted large language models for years. In late 2024 a preview of a model trained to spend far more computation at inference time reported private-set scores in the range of typical human performance, at costs orders of magnitude above ordinary use; a harder successor set was released in 2025 on which scores fell sharply again. The gap closes, then is redefined, and which of those moves is more informative remains unclear.The most-cited contamination-resistant test is [François Chollet](https://aiwiki.ai/wiki/francois_chollet)'s Abstraction and Reasoning Corpus, built around tasks requiring a novel rule to be inferred from a handful of examples.[^chollet2019] It resisted large language models for years. In late 2024 a preview of a model trained to spend far more computation at inference time reported private-set scores in the range of typical human performance, at costs orders of magnitude above ordinary use; a harder successor set was released in 2025 on which scores fell sharply again. The gap closes, then is redefined, and which of those moves is more informative remains unclear. ## The case for nearness## The case for nearness lines 66–78 → 66–781 unchanged line not shown
**Scaling.** Empirical relationships between model scale, data and loss have held over several orders of magnitude, and frontier training compute grew at roughly six-month doubling times through the 2010s and early 2020s.[^sevilla2022] If capability tracks loss and loss tracks compute, extrapolation implies continued rapid gains. The inferential structure is the one examined in [[accelerating-change]], and its best-known long-range form is [[ray-kurzweil]]'s curve-fitting. The addition of inference-time computation as a second scaling axis from 2024 onward reopened headroom that pretraining scaling alone appeared to be exhausting.**Scaling.** Empirical relationships between model scale, data and loss have held over several orders of magnitude, and frontier training compute grew at roughly six-month doubling times through the 2010s and early 2020s.[^sevilla2022] If capability tracks loss and loss tracks compute, extrapolation implies continued rapid gains. The inferential structure is the one examined in [[accelerating-change]], and its best-known long-range form is [[ray-kurzweil]]'s curve-fitting. The addition of inference-time computation as a second scaling axis from 2024 onward reopened headroom that pretraining scaling alone appeared to be exhausting. **Generality already observed.** A single model trained on next-token prediction can write code, translate, summarise clinical notes, prove competition mathematics and control a browser. That breadth was not designed in, and it is the strongest evidence against the older view that generality requires explicit architecture — the position Richard Sutton called the bitter lesson.**Generality already observed.** A single model trained on next-token prediction can write code, translate, summarise clinical notes, prove competition mathematics and control a browser. That breadth was not designed in, and it is the strongest evidence against the older view that generality requires explicit architecture, the view rejected by what Richard Sutton called the bitter lesson. **Expert opinion has moved.** Large surveys of published AI researchers have shifted median estimates for human-level machine intelligence substantially earlier across successive editions, with the aggregate 50% point falling around mid-century in the most recent.[^grace2024] Such forecasts have a poor track record in both directions, but the direction of movement is itself data.**Expert opinion has moved.** Large surveys of published AI researchers have shifted median estimates for human-level machine intelligence substantially earlier across successive editions, with the aggregate 50% point falling around mid-century in the most recent.[^grace2024] Such forecasts have a poor track record in both directions, but the direction of movement is itself data. ## The case against## The case against **Jaggedness.** Capability is not a scalar. Systems that solve olympiad problems fail at tasks a child handles, and the failures are not obviously less "general" than the successes. Studies of professional use find a frontier uneven in ways users cannot predict — a different situation from a uniformly sub-human system approaching parity.[^dellacqua2023]**Jaggedness.** Capability is not a scalar. Systems that solve olympiad problems fail at tasks a child handles, and the failures are not obviously less "general" than the successes. Studies of professional use find a frontier uneven in ways users cannot predict, which differs from a uniformly sub-human system approaching parity.[^dellacqua2023] **Reliability rather than capability is the binding constraint.** A system correct 95% of the time is not 95% of the way to automating a task if errors are uncorrelated with confidence and expensive to detect. Long-horizon agentic work compounds this: per-step reliability must be very high for hundred-step tasks to succeed.**Reliability rather than capability is the binding constraint.** A system correct 95% of the time is not 95% of the way to automating a task if errors are uncorrelated with confidence and expensive to detect. Long-horizon agentic work compounds this: per-step reliability must be very high for hundred-step tasks to succeed. lines 90–103 → 90–10311 unchanged lines not shown
Almost every long-horizon claim on this wiki has an AI term in it. Rejuvenation biology is bottlenecked on a system with more interacting variables than experiments can resolve, and [[ai-protein-design|protein structure prediction]] has already changed what is tractable there; companies such as [[retro-biosciences]] have made model-assisted protein engineering an explicit programme. The limit of that transfer is visible in [[ai-drug-discovery]], where models now generate candidate molecules far faster than medicinal chemists did and no drug produced that way has been approved. [[connectomics]] and [[whole-brain-emulation]] are limited by segmentation and simulation, both compute-bound. [[neural-decoding]] improves with better sequence models. Conversely, [[dual-use-research]] risk rises if design tools lower the expertise barrier for pathogens.Almost every long-horizon claim on this wiki has an AI term in it. Rejuvenation biology is bottlenecked on a system with more interacting variables than experiments can resolve, and [[ai-protein-design|protein structure prediction]] has already changed what is tractable there; companies such as [[retro-biosciences]] have made model-assisted protein engineering an explicit programme. The limit of that transfer is visible in [[ai-drug-discovery]], where models now generate candidate molecules far faster than medicinal chemists did and no drug produced that way has been approved. [[connectomics]] and [[whole-brain-emulation]] are limited by segmentation and simulation, both compute-bound. [[neural-decoding]] improves with better sequence models. Conversely, [[dual-use-research]] risk rises if design tools lower the expertise barrier for pathogens. The [[technological-singularity]] argument depends entirely on this article's subject: an intelligence explosion requires systems capable of improving themselves, which is a strictly stronger condition than AGI. [[intelligence-amplification]] and [[human-ai-merger]] describe the alternative in which capability grows through human–machine teams rather than autonomous systems, and [[mind-uploading]] would require both AGI-scale compute and neuroscience that does not exist. Whether such a system would be a subject as well as an agent is a separate question, treated in [[machine-consciousness]]; the [[substrate-independence]] premise it shares with most [[posthuman]] forecasting is assumed far more often than it is argued.The [[technological-singularity]] argument depends entirely on this article's subject: an intelligence explosion requires systems capable of improving themselves, which is a strictly stronger condition than AGI. [[intelligence-amplification]] and [[human-ai-merger]] describe the alternative in which capability grows through human-machine teams rather than autonomous systems, and [[mind-uploading]] would require both AGI-scale compute and neuroscience that does not exist. Whether such a system would be a subject as well as an agent is a separate question, treated in [[machine-consciousness]]; the [[substrate-independence]] premise it shares with most [[posthuman]] forecasting is assumed far more often than it is argued. > [!caution] Forecasts are not findings> [!caution] Forecasts are not findings> No timeline for AGI is well supported. Survey medians, scaling extrapolations and insider statements are all weak evidence, and the field's forecasting record — including confident predictions of imminence in the 1960s and of impossibility in the 1990s — should discount any specific date, including recent ones.> No timeline for AGI is well supported. Survey medians, scaling extrapolations and insider statements are all weak evidence, and the field's forecasting record, which includes confident predictions of imminence in the 1960s and of impossibility in the 1990s, should discount any specific date, including recent ones. ## Outlook## Outlook Governance has moved faster than definition. The EU's AI Act imposes obligations on general-purpose models by capability thresholds measured in training compute, several states have established safety institutes with evaluation mandates, and an international expert report on advanced AI risk was published in 2025. Every one of these instruments must operationalise "general" and "capable" without a scientific definition of either, and they have done so with proxies — training FLOP, benchmark scores, red-team results — that are known to be poor. Whether a capability threshold can be specified precisely enough to regulate before anyone can specify it precisely enough to measure is the practical form of the definitional problem, and it now binds.Governance has moved faster than definition. The EU's AI Act imposes obligations on general-purpose models by capability thresholds measured in training compute, several states have established safety institutes with evaluation mandates, and an international expert report on advanced AI risk was published in 2025. Every one of these instruments must operationalise "general" and "capable" without a scientific definition of either, and they have done so with proxies known to be poor: training FLOP, benchmark scores, red-team results. Whether a capability threshold can be specified precisely enough to regulate before anyone can specify it precisely enough to measure is the practical form of the definitional problem, and regulators now face it directly. ## See also## See also removed, struck through added, underlinedLine numbers count the serialised markdown of each revision, frontmatter included.
111 lines are the same in both revisions, and 58 of them are not shown. Neither revision here is necessarily the current article — read Artificial general intelligence for the text this wiki stands behind today.