🔍 Read the full analysis: How To Read 50 Signs Of Superintelligence In The AI Era on ThorstenMeyerAI.com
Prime made for students and young adults
- Fast, free delivery for dorm and study essentials
- Prime Video and Amazon Music included
- Member-only deals
TL;DR
A review of 18 claims in Alan D. Thompson’s 50-sign superintelligence tracker found seven verified, five based on vendor claims, three resting on a single source and three disputed or contradicted. The tracker marks two of 50 signs achieved, but its definition of superintelligence is closer to expert-level ability than some research labs’ definitions.
A review of 18 claims in researcher Alan D. Thompson’s 50-sign artificial superintelligence tracker found seven verified, five supported only by companies’ own claims, three based on a single source and three disputed or contradicted. The review, published October 26, finds credible AI advances in mathematics and science, but says the tracker’s broad definition and loose use of “partially achieved” should not be read as evidence that superintelligence has arrived.
Thompson, an Australian AI consultant who publishes The Memo newsletter, maintains a public checklist of 50 indicators, ranging from mathematical discoveries to technologies such as household robots. Its October 26 status grid lists two signs as achieved, 33 as partially achieved and 15 as not achieved. In an August correction note, Thompson said partial status includes funded programs, research infrastructure, formal scientific missions and other enabling work. That means the category can include early preparation rather than a near-complete capability.
The review examined selected claims from Thompson’s October video and tracker against papers, researchers’ accounts, company posts and regulatory filings. It reports stronger support for results with independently checkable evidence. Examples include AI contributions to mathematics and a result on a unit-distance conjecture described by OpenAI and examined by external mathematicians. Other entries, including claims about a model’s performance on mathematical problems, rely on company-reported results or have qualifications about human guidance, problem formulation or prior literature.
Among the caveats, mathematicians Terence Tao, Don Knuth and Tim Gowers are cited as describing genuine AI contributions while also explaining limits on how the work was done. The review also highlights a gap in claims about AI improving its own development: Anthropic’s reported internal evaluation puts its best model at 62.8% on a test the company says would require at least 85% for a full replacement for its researchers. Anthropic estimates AI makes its research and development faster, but “not yet by a factor of 2.”
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
The Gap Between Claims and Capability
The tracker draws attention to developments that could matter, but its status labels combine very different kinds of evidence. A proven mathematical result, a company benchmark and a funded research program do not establish the same level of capability. Treating all three as steps toward an achieved sign can make progress appear more settled than it is.
That distinction matters for readers, policymakers and businesses trying to judge whether systems can perform reliably across fields, work autonomously or replace human expertise. The review finds that some specific advances stand up to scrutiny, while broader conclusions about superintelligence remain dependent on definitions and evidence that are not consistent across the tracker and AI labs.
Popular right now
Ninja Luxe Café Premier 3-in-1 Espresso Machine, Drip Coffee & Cold Brew
De'Longhi Linea Classic Espresso Machine with Adjustable Milk Frother
Vitamix Propel Series 750 Professional-Grade Blender, 4 Programs, Black
As an affiliate, we earn on qualifying purchases.
Different Definitions Set Different Bars
Thompson defines artificial general intelligence as a machine performing at the level of an average human and artificial superintelligence as a machine performing at the level of an expert human in practically any field. Under that definition, an expert-level result in one area can be relevant to his ASI checklist, even if the system has not shown comparable ability across other domains.
Google DeepMind’s 2023 “Levels of AGI” framework separates competent, expert, virtuoso and superhuman performance. Its June 2026 report uses a more demanding description of superintelligence: systems more capable than large organisations of humans. OpenAI’s charter, meanwhile, defines AGI around highly autonomous systems that outperform humans at most economically valuable work. These frameworks differ on breadth, autonomy and the human comparison group.
Thompson’s separate “conservative countdown to AGI” stood at 99% in the source material. He said the remaining step depended on a robot completing multi-step fine assembly, such as assembling flat-pack furniture, with average-human reliability. The figure is his estimate, not a consensus measurement. The review also says household robots have not appeared at scale, there is no verified AI-run company worth a billion dollars, and no visible break in the US unemployment rate attributable to AI.
“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 about the tracker
What the Tracker Cannot Establish
The review covers 18 selected claims, not all 50 indicators, and its findings do not amount to a comprehensive independent audit of the full checklist. The source material gives categories for all 18—seven verified, five vendor claims, three single-source and three disputed or contradicted—but the supplied account does not provide the full claim-by-claim evidence for every entry.
“Verified” also applies to particular claims, not to the broader conclusion that AI has reached superintelligence. Some mathematical work involved human feedback or coaching, and company performance figures may not have been reproduced independently. The tracker’s definitions and status rules are Thompson’s own; the source material does not establish that researchers or AI labs have adopted them as a shared standard.
It remains unclear when, or whether, systems will meet broader tests of autonomy and dependable performance across fields. The review also does not establish that AI has caused, or has not caused, wider economic changes such as shifts in employment.
Evidence to Watch in Future Updates
The next useful test is whether claimed results can be independently reproduced and whether new capabilities work without extensive coaching or specially chosen tasks. Readers following the tracker can also compare later status changes with its stated criteria, particularly for entries marked partially achieved.
No specific next assessment date is given in the source material. Until further evidence appears, the review’s practical conclusion is to treat the checklist as one researcher’s monitoring framework, not a measurement that settles whether superintelligence has arrived.
Key Questions
Has the review concluded that artificial superintelligence has arrived?
No. It found evidence for some specific AI advances, but says the tracker’s definition is not shared by all labs and does not establish that systems exceed the capabilities of large human organisations.
How many signs in Thompson’s tracker are marked achieved?
The tracker’s status grid, last updated October 26, lists two achieved, 33 partially achieved and 15 not achieved. Thompson’s definition of partial progress includes funded programs and enabling work.
What did the review find among the 18 claims it checked?
It classified seven as verified, five as vendor claims, three as based on a single source and three as disputed or contradicted.
Why do definitions of superintelligence matter?
Thompson’s definition focuses on expert-human performance across fields. Google DeepMind’s 2026 description sets a higher bar, comparing a system with large organisations of humans. A claim can therefore fit one framework without meeting another.
Source: ThorstenMeyerAI.com
