🔍 Read the full analysis: 50 Superintelligence Signs: A Framework For Thinking About AGI And ASI on ThorstenMeyerAI.com
Get the latest gadgets delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
AI consultant Alan D. Thompson’s 50-indicator tracker, updated Oct. 26, 2026, marks two signs achieved, 33 partially achieved and 15 not achieved. A review of 18 prominent claims found seven verified, five based on vendor reports, three supported by a single source and three disputed or contradicted; the results depend in part on how “superintelligence” is defined.
A review of 18 prominent claims in AI consultant Alan D. Thompson’s 2026 tracker of signs of artificial superintelligence found seven verified, five resting on company claims, three supported by a single source and three disputed or contradicted. Thompson’s checklist, updated Oct. 26, labels two of its 50 indicators achieved, 33 partially achieved and 15 not achieved, but its broad definition of superintelligence differs from those used by major AI labs.
Thompson, an Australian AI consultant who publishes The Memo newsletter, lists indicators ranging from mathematical discoveries and recursive code improvement to planetary exploration and faster-than-light travel. His October video, “ASI: A better world – 2026,” presents developments from January through October. He describes Claude Mythos Preview and GPT-6 Astra as “early ASI,” a characterization that reflects his own framework rather than an agreed industry designation.
The tracker’s status categories need qualification. In an August correction note, Thompson said “partially achieved” covers “documented material progress” including funded programmes, research infrastructure, formal scientific missions and enabling work. That means the category does not necessarily indicate that a system is close to meeting the indicator. The source review, published by Thorsten Meyer on Oct. 26, also notes that some examples placed under an indicator do not demonstrate its literal endpoint.
Evidence was strongest for claims that could be checked against public work or independent expertise, including reported contributions to mathematics and an AI-related physics calculation. The review says mathematicians Terence Tao, Donald Knuth and Timothy Gowers have described genuine AI contributions, while noting limits such as human guidance, existing literature and the nature of the problems. Other claims rely on company-reported results or have not been independently reproduced.
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
Why the Evidence Standards Matter
The tracker offers readers a way to follow developments across research, automation and science, but its traffic-light scores can be misleading if treated as a measured countdown to superintelligence. “Partially achieved” includes enabling work, not just a system demonstrating the stated capability. And a finding verified in one narrow domain does not establish broad competence across fields.
Definitions also change the significance of a claim. Thompson defines ASI as performance at expert-human level in practically any field. Google DeepMind’s 2023 framework calls expert-level performance “Expert AGI,” while its June 2026 report describes superintelligence as exceeding the capabilities of large organisations of humans. OpenAI’s charter, meanwhile, defines AGI around highly autonomous systems that outperform people at most economically valuable work. These measures emphasize different things: skill level, scale of capability and autonomy.
The review points to a further gap between progress in software and real-world deployment. It says there are not yet household robots operating at scale, a verifiably billion-dollar AI-run company, or a visible break in the US unemployment rate attributable to AI. Those observations do not settle what future systems may do, but they caution against treating technical demonstrations as proof of economy-wide transformation.
Popular right now
Philips 4400 Series Fully Automatic Espresso Machine with LatteGo Milk System
De'Longhi Stilosa Compact Manual Espresso Machine with Milk Frother
TCL 85 Designer Series QD-Mini LED TV (Model 85A400U) with Google Smart TV
As an affiliate, we earn on qualifying purchases.
Three Different Yardsticks for ASI
Thompson’s checklist is a public collection of 50 indicators rather than a standardized scientific test. His definitions put average-human performance at AGI and expert-human performance across practically any field at ASI. Under Google DeepMind’s 2023 scale, “Expert” is the 90th percentile of skilled adults; the framework also distinguishes “Virtuoso” and “Superhuman” levels. The labels therefore do not map neatly onto one another.
The review also separates evidence into five verdict types: verified, vendor claim, single source, disputed or contradicted, and not found. Its examples include mathematical work associated with GPT-5.2, Claude Opus 4.6 and an OpenAI model. The review reports external verification for some mathematical claims, while stressing that the accounts include caveats: one result followed feedback, another involved coaching and a proof written by Knuth, and a company-reported set of formalized results remains a vendor claim.
Thompson maintains a separate “conservative countdown to AGI” at 99%. He says the remaining step depends on a robot completing multi-step fine assembly, such as putting together furniture, with average-human reliability. That countdown is another judgment within his framework, not an independently measured probability that AGI is imminent.
““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
What the Tracker Cannot Establish
The review’s 18-claim sample is not a complete audit of all 50 indicators, and its findings should not be read as a definitive score for the field. The supplied account does not provide the full evidence trail and verdict for every item. Some results are independently checked; others are company claims, single-source reports or contested accounts.
It also remains unsettled whether systems called “early ASI” by Thompson meet another researcher’s threshold for AGI or superintelligence. The categories and definitions are not standardized, and capability demonstrated on a math problem does not by itself establish autonomous, general performance. The review’s claims about unemployment, household robotics and AI-run companies likewise describe the present evidence as reported; they do not show what may change later.
Further Checks on AI Claims
The next useful steps are independent replication of vendor-reported results, clearer disclosure of human input and evaluation methods, and updates to the tracker as new evidence appears. Readers can compare future entries with primary papers, lab statements and official filings rather than relying on a status color or a short video montage.
Thompson’s tracker was current as of October 26, 2026, according to the source material. Whether any indicator moves from partial to achieved will depend on the precise capability being tested, the quality of the evidence and which definition of AGI or ASI is being applied.
Key Questions
How many signs does Thompson’s tracker mark as achieved?
As of October 26, 2026, it marks two of 50 indicators achieved, 33 partially achieved and 15 not achieved.
Does “partially achieved” mean superintelligence is nearly here?
No. Thompson says the category includes funded programmes, research infrastructure and enabling work, so it does not necessarily mean the system has nearly met the indicator.
Were the AI research claims independently verified?
Some were. The review classified seven of 18 sampled claims as verified, five as vendor claims, three as single-source and three as disputed or contradicted.
Do AI researchers use the same definition of superintelligence?
No. Thompson’s expert-human-level definition differs from Google DeepMind’s description of systems exceeding large human organisations and from other frameworks focused on autonomy or economic performance.
Source: ThorstenMeyerAI.com
