🔍 Read the full analysis: 50 Signs That Frame The Debate Over AI, AGI, And ASI on ThorstenMeyerAI.com
Get privacy and security gear 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
Thorsten Meyer’s October 26 review of researcher Alan D. Thompson’s 50-sign ASI tracker says 2 indicators are marked achieved, 33 partially achieved and 15 not achieved. Among 18 prominent claims Meyer checked, 7 were verified, 5 were vendor claims, 3 relied on a single source and 3 were disputed or contradicted; the review also stresses that “early ASI” depends on whose definition is used.
Thorsten Meyer’s October 26 review of a 50-sign artificial superintelligence tracker found that its strongest claims have mixed levels of evidence, while the tracker itself lists 2 indicators as achieved, 33 as partially achieved and 15 as not achieved. The review checks 18 prominent claims against research papers, researcher accounts, company statements and filings, and says the results do not establish that today’s leading AI systems are superintelligent.
The tracker is published by Dr. Alan D. Thompson, an Australian AI consultant who describes systems such as Claude Mythos Preview and GPT-6 Astra as “early ASI.” Meyer reports that the tracker’s status grid, last updated October 26, marks 2 of 50 indicators achieved. Its “partially achieved” category includes funded programmes, research infrastructure and enabling work, so it does not mean a system is close to meeting the indicator.
Meyer examined 18 selected claims from Thompson’s video and tracker. His classifications were 7 verified, 5 vendor claims, 3 based on a single source, and 3 disputed or contradicted. The review says independently checkable results cluster around mathematical work, a physics calculation and regulatory filings. Its labels describe Meyer’s source review, not a formal scientific assessment of whether ASI has arrived.
Among the examples Meyer calls verified, mathematician Terence Tao described an AI contribution to an Erdős problem as “more or less autonomously,” while noting that feedback was involved, the problem had been misformulated and related results existed. Donald Knuth’s account of a separate problem said Claude found a construction after coached exploration; Knuth wrote the proof. Meyer also reports external mathematicians verified an AI-assisted result concerning an Erdős conjecture. These accounts support meaningful contributions, but do not show independent, general-purpose expert performance across fields.
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 ASI label changes
The debate matters because a label such as “early ASI” can imply a capability threshold that researchers and companies do not define in the same way. Thompson defines ASI as performance at the level of an expert human in practically any field. In Google DeepMind’s 2023 scale, that roughly matches “Expert” AGI, while its “Superhuman” level is above all human performance. DeepMind’s June 2026 report uses a more demanding description of superintelligence: systems more capable than large organisations of humans.
OpenAI’s charter, as summarized in Meyer’s review, defines AGI around highly autonomous systems that outperform people at most economically valuable work. That emphasis on autonomy differs from a system producing a strong result with expert coaching or human-written proof work. For readers, the distinction affects how claims about jobs, research and near-term risks should be interpreted: a domain-specific achievement is evidence of progress, but not by itself evidence of broadly autonomous intelligence.
The tracker also combines very different kinds of indicators, from mathematical results to household robotics and speculative milestones. A status of “partial” can refer to preparatory work rather than a demonstrated capability. Meyer says no household robots are operating at scale, no AI-run company has been verifiably shown to be worth a billion dollars, and there is no visible break in the U.S. unemployment rate attributable to AI in the material reviewed.
Products Worth Considering
Logitech MX Mechanical Wireless Illuminated Keyboard, Tactile Quiet, Graphite
Amazon Kindle Scribe Colorsoft 64GB (1st Generation)
Samsung 27-Inch Odyssey QD-OLED G8 4K 240Hz Gaming Monitor
As an affiliate, we earn on qualifying purchases.
How the tracker defines progress
Thompson’s public checklist contains 50 proposed indicators, ranging from recursive code self-optimization to much more speculative entries. His August correction note says “partial” includes documented progress toward an indicator, including funded programmes, research infrastructure, formal scientific missions and enabling work. That makes the category broader than demonstrations of the end capability.
The review gives the example of two exoplanet candidates reportedly found with Claude Code being listed under an indicator about the first planet other than Earth being colonized. Meyer says this illustrates how an item can be assigned partial status because of related work, even when the stated milestone has not occurred. Thompson also maintains a separate “conservative countdown to AGI” at 99%, with the remaining test described as a robot completing multi-step fine assembly at average-human reliability. These are Thompson’s assessments, not independently established measures of progress.
For its claim checks, Meyer distinguishes independent verification from a company’s own reporting. For example, the review calls some mathematics results verified when outside experts or published work support them, while treating other performance figures as vendor claims when they have not been independently reproduced. That difference is central to reading a fast-moving field where public demonstrations, lab evaluations and real-world autonomy are not interchangeable.
““More or less autonomously””
— Terence Tao, as quoted in Meyer’s review
What the evidence cannot establish
The review covers 18 selected claims, not all 50 tracker entries, and its classifications depend on the sources available to Meyer. The supplied account does not provide the full evidence behind every claim or enough detail to independently reproduce the review’s checks. Five claims are vendor claims, three rely on a single source, and three are disputed or contradicted, leaving important questions about replication and interpretation.
There is also no shared definition of AGI or ASI across the tracker and the cited labs. A mathematical result, even when independently supported, does not establish expert-level ability in practically any field or the autonomy required by other definitions. The review’s statements about employment and commercial value likewise do not prove that AI has had no effects; they say the cited indicators have not visibly or verifiably crossed those particular thresholds.
Evidence to watch in future updates
The next useful developments will be independent replication of claimed results, clearer reporting on how much human guidance was involved, and evaluations that test systems across fields rather than on selected tasks. For robotics, the tracker’s own proposed benchmark is reliable multi-step fine assembly in ordinary settings. Progress on that kind of task would address a different capability gap from success on mathematical problems.
Thompson’s tracker may change as new results appear, while labs may publish updated evaluations and definitions. Readers should compare each new claim with the precise milestone, its baseline and the source of the evidence. The current review supports a picture of notable, uneven progress; it does not settle whether AGI or ASI has arrived.
Key Questions
What did the review find about the 50-sign tracker?
As of October 26, 2026, the tracker marked 2 indicators achieved, 33 partially achieved and 15 not achieved. Its partial category includes enabling work and funded programmes, not only demonstrated end capabilities.
Did the review conclude that ASI has arrived?
No. It reports that Thompson calls some leading models “early ASI,” but explains that definitions differ and the evidence reviewed does not establish broad, autonomous superintelligence.
How many of the 18 claims were independently verified?
Meyer classified 7 as verified, 5 as vendor claims, 3 as based on a single source and 3 as disputed or contradicted. These are the review’s classifications of selected claims.
Why do definitions of ASI matter?
Thompson’s definition centers on expert-human performance across fields, while DeepMind and OpenAI use different thresholds involving superhuman capability or autonomy. A claim can appear to meet one definition without meeting another.
Source: ThorstenMeyerAI.com
