🔍 Read the full analysis: 50 Signs Of Superintelligence, Checked: How Close Is AI To 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
Thorsten Meyer’s October 26 review checked 18 prominent claims from researcher Alan D. Thompson’s 50-indicator superintelligence tracker. It classifies seven as verified, five as vendor claims, three as based on a single source and three as disputed or contradicted; the tracker itself marks two of 50 indicators achieved. The review finds credible advances in mathematics and science, but no evidence that AI has reached a broadly agreed definition of artificial superintelligence.
A review published by ThorstenMeyerAI.com on October 26 checked 18 prominent claims in researcher Alan D. Thompson’s list of 50 signs of artificial superintelligence, finding seven verified, five based on company claims, three resting on a single source and three disputed or contradicted. Thompson’s tracker marks two indicators achieved, but the review says its broad definition of “partially achieved” and the varying meanings of AGI and ASI make the checklist an opinionated watchlist, not a measure showing that superintelligence has arrived.
Thompson, an Australian AI consultant who publishes The Memo newsletter, describes AGI as a machine performing at the level of an average human and ASI as a machine performing at the level of an expert human in practically any field. His public 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” can include funded programs, research infrastructure, scientific missions and other enabling work. That category, the review cautions, does not mean a system is close to completing the listed milestone.
Meyer examined 18 headline claims and checked them against researchers’ accounts, papers, company publications and other primary material. The review says the clearest independently checked examples involve mathematical work. Mathematicians Terence Tao, Don Knuth and Tim Gowers have described AI contributions, but with qualifications about human guidance, existing literature and the kinds of problems involved. The review also reports an externally checked result involving an OpenAI model and a disproof of a mathematical conjecture. Other claims, including results announced by AI companies, are classified as vendor claims where the figures have not been independently reproduced.
Thompson’s October video describes Claude Mythos Preview and GPT-6 Astra as “early ASI.” That wording is Thompson’s assessment, not a consensus finding established by the review. The tracker also lists two exoplanet candidates identified with Claude Code under an indicator about colonizing a planet other than Earth. Meyer presents that classification as an example of how the tracker can count enabling work toward a distant milestone, rather than evidence that colonization has occurred.
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 scorecard matters
The disagreement is not only about how capable the newest models are. It is also about what counts as AGI or ASI, and whether a reported achievement demonstrates broad, reliable performance or a narrow success with human assistance. A checklist that groups research funding, tools and completed capabilities under “partially achieved” can help readers follow developments, but its color-coded status should not be read as a probability or a countdown unless its author defines it that way.
That distinction matters to policymakers, businesses and the public as AI systems enter research and other skilled work. Verified advances are meaningful, but they do not by themselves establish that systems can act autonomously across fields, replace expert teams or carry out complex physical tasks reliably. The review reports no household robots operating at scale, no verifiably billion-dollar AI-run company and no visible break in the US unemployment rate attributed to AI. Those observations do not settle AI’s future economic effects; they show that claims about present-day superintelligence should not be inferred from a handful of technical results.
The review also highlights evidence about AI-assisted AI research. Anthropic’s own reported evaluation gives its best internal model a 62.8% score on a test the company says would require at least 85% for a full replacement for its researchers. Anthropic estimates that AI makes its research and development faster, but “not yet by a factor of 2.” These are company-reported figures, not an independent measurement of the entire field, and they point to a more limited claim than fully autonomous, self-improving research.
Products Worth Considering
Jackery Explorer 300 Portable Power Station, 292Wh
Logitech MX Keys S Wireless Keyboard, Graphite
Apple iPad Air (4th Generation, 10.9-inch, Wi-Fi, 256GB) – Space Gray
As an affiliate, we earn on qualifying purchases.
Different scales, different thresholds
Thompson’s definitions are not the only ones in use. Google DeepMind’s 2023 “Levels of AGI” framework distinguishes performance levels from competent through expert and virtuoso to superhuman. Under that scale, expert performance is not the same as superhuman performance. Meyer says Thompson’s definition of ASI is roughly comparable to DeepMind’s “Expert AGI,” rather than the framework’s highest level.
Google DeepMind’s June 2026 report, “From AGI to ASI,” describes superintelligence as a system more capable than large organisations of humans. OpenAI’s charter uses a different AGI threshold, focusing on highly autonomous systems that outperform humans at most economically valuable work. The definitions put different weight on expert skill, broad superiority and autonomy. A model producing an expert-level result in one area therefore does not automatically meet every organisation’s threshold for AGI or ASI.
Thompson separately maintains a “conservative countdown to AGI,” which he reports at 99%. He says the remaining gap is a robot able to perform multi-step fine assembly, such as putting together flat-pack furniture, with average-human reliability. That is his own countdown and chosen benchmark, not an agreed industry measure. The distinction helps explain why a high percentage on one tracker cannot be treated as a field-wide estimate of how close AI is to AGI.
“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 his tracker
What the tracker cannot establish
The review covers 18 selected claims, not all 50 indicators, and its classifications depend on the evidence available for each claim. The supplied account does not give a complete breakdown of every item or establish a shared, independently administered test for superintelligence. Company-reported benchmarks may be useful evidence, but the review labels several as vendor claims because independent replication is lacking.
It also remains unclear how well success on specific research problems generalises to reliable performance across fields, or how much human input is needed for each result. The tracker’s label “early ASI” is Thompson’s interpretation. No broad agreement that today’s leading models qualify as ASI is established by the material reviewed.
Evidence to watch next
The next useful milestones are results that can be independently checked: further mathematical work with clear accounts of human involvement, reproducible benchmarks, and demonstrations of capable systems performing tasks across domains. For claims about autonomy, evidence would need to show not only that a model can solve a problem, but that it can do so reliably with limited supervision.
Readers can also watch for updates to Thompson’s tracker and whether its definitions or completion criteria change. Until the listed milestones are supported by stronger, comparable evidence, the most accurate reading is that AI has made documented advances in particular areas, while the claim that superintelligence has arrived remains unsettled.
Key Questions
How many of the 50 ASI indicators are marked achieved?
Thompson’s tracker, reported as updated on October 26, 2026, marks two achieved, 33 partially achieved and 15 not achieved. The review notes that “partially” includes enabling work and funded programs.
Does the review say that AI has reached ASI?
No. It reports that Thompson calls two models “early ASI,” but presents that as his assessment. The review does not establish a consensus that current AI systems meet a broadly accepted ASI definition.
What did the review find about AI and mathematics?
It describes several mathematical contributions as supported by researchers or external experts, while noting caveats involving coaching, problem formulation and prior literature. It separates those cases from company claims not independently reproduced.
Why do AGI and ASI definitions matter?
Different frameworks set different bars for expert-level performance, broad superiority and autonomy. A model’s success on a specific task may meet one definition or level without establishing general or superhuman capability.
Source: ThorstenMeyerAI.com
