🔍 Read the full analysis: 50 Signs Checked: What They Mean For AI, AGI, And ASI on ThorstenMeyerAI.com
Get business pricing on office and shipping supplies
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
AI consultant Alan D. Thompson’s 50-item ASI tracker lists two signs as achieved, 33 as partially achieved and 15 as not achieved. A review of 18 prominent claims found seven verified, five based on vendor claims, three resting on a single source and three disputed or contradicted; definitions of superintelligence also differ sharply.
A review of 18 prominent claims in AI consultant Alan D. Thompson’s 50-sign artificial superintelligence tracker found that seven were verified, while others relied on company statements, single sources or disputed evidence. Thompson’s tracker marks just two of 50 indicators as achieved; its broad “partially achieved” category covers work in progress, not proof that superintelligence is near.
Thompson, an Australian AI consultant and publisher of The Memo newsletter, maintains a public checklist of signs that he believes could indicate the arrival of artificial superintelligence, or ASI. Its entries range from AI-assisted mathematical discoveries to speculative milestones such as faster-than-light travel. In a video published in October, he presented developments from January through October 2026 and described Claude Mythos Preview and GPT-6 Astra as “early ASI.” That characterization is Thompson’s assessment, not a consensus finding or a measurement accepted across the field.
The tracker’s October 26 status grid lists two indicators as achieved, 33 as partially achieved and 15 as not achieved. In an August correction note, Thompson said “partially achieved” includes “documented material progress” such as funded programmes, research infrastructure, formal scientific missions and enabling work. The category can therefore cover early or indirect activity; it does not mean an indicator is almost complete. For example, the tracker places two exoplanet candidates identified with Claude Code under an indicator about the first planet beyond Earth being colonized.
The review’s findings were uneven across the claims examined. Its verified examples cluster in areas where results can be checked, including mathematical work, a physics calculation and regulatory filings. Claims about a series of decade-old mathematical results attributed to GPT-6 Astra, for instance, were classed as vendor claims in the source review: Lean certificates were public, but people had prepared the problems. The review also says some claimed AI contributions to mathematics had meaningful qualifications, including human coaching, prior literature and the style of the problems involved.
Whose superintelligence?
One word, very different bars
Dr Alan D. Thompson's ASI starts where Google DeepMind's scale puts "Expert AGI". The same model can be "early ASI" on one scale and a strong but narrow AGI candidate on another.
- Median human50th percentileThompson: AGIa machine at the level of an average (median) humanDeepMind: Competent AGIat least the 50th percentile of skilled adults
- Expert90th percentileThompson: ASIexpert-human level in practically any fieldDeepMind: Expert AGIat least the 90th percentile of skilled adults
- Virtuoso99th percentile—DeepMind: Virtuoso AGIat least the 99th percentile of skilled adults
- Beyond every human100%—DeepMind: Superhuman (Level 5)outperforms 100% of humans
- Beyond organisations2026 report—DeepMind: ASImore capable than large organisations of humans
Thompson’s tracker status as of 26 October 2026. DeepMind rows: “Levels of AGI” (2023) and, for the last row, “From AGI to ASI” (June 2026). OpenAI’s Charter: AGI = “highly autonomous systems that outperform humans at most economically valuable work”.
Sources: LifeArchitect.ai ASI checklist (26 Oct 2026), AGI countdown, Morris et al., Levels of AGI (2023), Genewein et al., From AGI to ASI (2026), OpenAI Charter.
Powered by Thorsten Meyer AI
Evidence Behind the ASI Claims
The tracker gathers developments that can otherwise be hard to compare, but its labels combine very different kinds of evidence. A verified mathematical result, a company’s performance figure, a funded research effort and a speculative milestone do not establish the same thing. Readers should treat the checklist as one researcher’s framework for tracking progress, rather than a standardized test showing that ASI has arrived.
Definitions also change what counts as progress. Thompson defines ASI as a machine performing at the level of an expert human in practically any field. Google DeepMind’s 2023 capability scale calls the 90th-percentile level “Expert” and places “Virtuoso” and “Superhuman” above it. DeepMind’s June 2026 report uses a more demanding description of superintelligence: systems more capable than large organisations of humans. OpenAI’s charter, meanwhile, describes AGI through highly autonomous systems that outperform humans at most economically valuable work. These are different benchmarks, not interchangeable labels.
The distinction matters for public discussion and decisions about AI readiness. A model that contributes to an expert-level proof may be important without demonstrating broad, reliable autonomy across fields. The review also points to gaps between advances in software and claims about wider economic or physical effects: it reports no household robots at scale, no verifiably billion-dollar AI-run company and no visible break in the US unemployment rate tied to AI. Those observations do not settle what comes next, but they caution against treating technical milestones as evidence of economy-wide transformation.
Products Worth Considering
Samsung 27-Inch Odyssey QD-OLED G8 4K 240Hz Gaming Monitor
FLEXISPOT 42-Inch Electric Standing Desk Converter
FLEXISPOT 32-Inch Electric Standing Desk Converter with Keyboard Tray
As an affiliate, we earn on qualifying purchases.
How the Tracker Defines Progress
Thompson’s checklist is one of several ways researchers and commentators frame progress toward advanced AI. His separate “conservative countdown to AGI” stood at 99% in the source report. He said the remaining percentage depended on a robot completing multi-step fine assembly, such as building IKEA furniture, at average-human reliability. That countdown uses a different threshold from the 50-sign ASI list, so the figures should not be read as a single shared measure of capability.
For the 18-claim review, the author says he traced assertions to primary materials such as papers, researcher notes, company posts and filings, then looked for critical responses. The review used categories including verified, vendor claim, single source, disputed or contradicted, and not found. Among the cited mathematical examples, Terence Tao described an AI contribution to an Erdős problem as made “more or less autonomously,” but after feedback and with a problem that had been misformulated; Don Knuth described Claude finding a construction after coached exploration and said he wrote the proof. These details show why human involvement and the exact claim matter when describing AI results.
““documented material progress toward an indicator, including funded programs, research infrastructure, formal scientific missions, and enabling work””
— Alan D. Thompson, in an August correction note on his tracker
Where the Evidence Falls Short
The review does not establish that any current system meets a broadly accepted definition of ASI. The term has no single agreed threshold in the material examined, and Thompson’s definition is less demanding than DeepMind’s description of systems exceeding the cognitive capability of large human organisations. Calling specific models “early ASI” should therefore be understood as Thompson’s interpretation, not an independently confirmed status.
The review’s classifications also have limits. The source reports that seven claims were verified, five were vendor claims, three rested on a single source and three were disputed or contradicted, but it does not provide enough detail here to independently assess every item or identify the status of all 18 claims. Company performance figures and results involving human-prepared tasks may require further independent replication. It is also unclear how the checklist’s status assignments will change as more evidence appears.
What Future Checks Need to Show
The next useful test is whether the reported capabilities can be reproduced and whether systems can perform them reliably with less human direction. For mathematical and scientific claims, that means examining the underlying work, the role of researchers and the extent to which results have been independently checked. For claims about AI improving AI research, performance on relevant tests and measured effects on research productivity will matter more than labels alone.
Thompson’s tracker is dated October 26, 2026, and future revisions may alter the status of individual signs. Until then, readers should distinguish demonstrated results from vendor statements, and both from indicators that record funding or enabling work. The source review does not identify a confirmed date when any broader ASI threshold will be met.
Key Questions
How many of the 50 signs does the tracker list as achieved?
As of October 26, 2026, Thompson’s tracker lists two as achieved, 33 as partially achieved and 15 as not achieved.
Does “partially achieved” mean a sign is almost complete?
No. Thompson’s definition includes funded programmes, research infrastructure and other enabling work. The label records progress, but does not indicate that a milestone is nearly complete.
Do the reviewed claims show that ASI has arrived?
No. The review checked 18 claims and found a mix of independently verified results, vendor claims, single-source reports and disputed or contradicted claims. It does not establish that a current AI system meets an agreed definition of artificial superintelligence.
Why do different sources use different definitions of ASI?
There is no single threshold in the source material. Thompson’s definition focuses on expert-human performance across practically any field, while DeepMind’s 2026 report describes systems more capable than large organisations of humans. The definition changes what counts as evidence.
Source: ThorstenMeyerAI.com
