🔍 Read the full analysis: What Counts As A Sign Of Superintelligence? 50 Clues About AI And AGI on ThorstenMeyerAI.com
Get smart everyday buys 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
An October 26 review of Dr. Alan D. Thompson’s 50-item superintelligence tracker says it marks two indicators achieved, 33 partially achieved and 15 not achieved. The review checked 18 headline claims, classifying seven as verified, five as vendor claims, three as single-source and three as disputed or contradicted. The results show meaningful AI progress, especially in mathematics, but do not establish that superintelligence has arrived.
A review published October 26 by ThorstenMeyerAI.com examined 18 prominent claims in AI consultant Dr. Alan D. Thompson’s 50-sign superintelligence tracker, finding seven supported by independent experts or official records, five based on company claims, three resting on a single source and three disputed or contradicted. Thompson’s tracker lists two indicators as achieved, 33 as partial and 15 as not achieved; those labels are his assessment, not evidence that artificial superintelligence has arrived.
The review describes the tracker as a collection of possible indicators, ranging from mathematical discoveries to robot capability and even speculative milestones. Its status grid, last updated October 26, marks 33 of 50 indicators as partial. Thompson has said that this category includes funded programs, research infrastructure and other enabling work, so a partial rating does not mean a system is close to meeting the full indicator.
The review’s examples show why the underlying evidence matters. It classifies reported AI contributions to mathematics as verified when researchers or independent mathematicians documented them, while distinguishing those results from broader claims about a model’s capabilities. For example, mathematician Terence Tao described one result as achieved “more or less autonomously” after feedback; the problem had been misformulated, and similar results already existed. In another case, Donald Knuth wrote that Claude found a construction during coached exploration, while Knuth supplied the proof.
Other headline items received less conclusive labels. The review calls a claim about a model solving ten decade-old problems for about $2,000 a vendor claim, even though Lean certificates were made public; it says human involvement in preparing the results matters when judging what the model did independently. The supplied review also points to Anthropic’s internal assessment of AI-assisted research and development: its best internal model scored 62.8% on a test for which Anthropic said a full substitute for its researchers would need at least 85%. The company estimated that 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
Why the labels matter
The tracker draws attention to concrete questions about what AI can do, but its categories combine unlike evidence. A published proof, a company-reported benchmark, a funded program and a speculative future milestone do not offer the same level of confirmation. Treating all orange “partial” entries as near-completions would overstate what the tracker says.
Definitions also change the conclusion. Thompson defines artificial superintelligence as performance at the level of an expert human in practically any field. Google DeepMind’s 2023 framework calls expert-level performance “Expert AGI,” while its 2026 report describes superintelligence as exceeding the cognitive capabilities of large human organisations. OpenAI’s charter, meanwhile, defines AGI around highly autonomous systems that outperform people at most economically valuable work. Whether a system counts as superintelligent therefore depends partly on the benchmark and definition being used, not just on a striking result in one domain.
For readers, the distinction is practical: mathematical or scientific progress can be real without showing broad autonomy, reliable performance across fields or major changes to employment. The review says it found no household robots operating at scale, no verifiably billion-dollar AI-run company and no visible break in the U.S. unemployment rate tied to the claims it examined. Those observations do not settle what may happen later; they caution against presenting isolated demonstrations as proof of economy-wide transformation.
Popular right now
Philips 4400 Series Fully Automatic Espresso Machine with LatteGo Milk System
Vitamix 5200 Professional-Grade Blender, 64 oz. Container, Black
TCL 85-Inch QM6K Series QLED Mini LED 4K HDR Smart Google TV
As an affiliate, we earn on qualifying purchases.
How Thompson defines the milestones
Thompson is an Australian AI consultant who publishes The Memo newsletter and maintains the public checklist. The tracker includes 50 proposed signs, from recursive code self-optimization to claims that are far more speculative. In an October video montage, he described Anthropic’s Claude Mythos Preview and OpenAI’s GPT-6 Astra as “early ASI.” That description is Thompson’s characterization; it is not an independently established classification in the review.
His scale differs from other published frameworks. Thompson defines AGI as a machine performing at the level of an average human and ASI as expert-human performance in practically any field. DeepMind’s 2023 scale separates “Competent,” “Expert,” “Virtuoso” and “Superhuman” performance. Its 2026 report uses a higher bar for superintelligence: capability beyond large organisations of humans. The review argues that this gap should be made explicit whenever labels such as “early ASI” or “proto-ASI” are used.
Thompson also maintains a separate, conservative countdown to AGI that he places at 99%. He says the remaining step depends on a robot completing multi-step fine assembly, such as assembling flat-pack furniture, with average-human reliability. That percentage is a feature of his own tracker, not a measured consensus among researchers.
“more or less autonomously”
— Terence Tao, as quoted in the review
What the audit cannot establish
The review checked 18 selected claims, not every entry in the 50-item tracker, and its findings should not be treated as a full independent evaluation of every AI system. The source material does not provide enough detail to assess all 18 items individually, including the complete evidence behind every disputed or single-source classification. Seven verified claims do not amount to verification of the tracker as a whole.
It also remains unclear how much current systems can perform without human selection, coaching, tool setup or correction across a wide range of tasks. Public demonstrations and company-reported benchmarks can document useful capabilities, but they do not by themselves show reliable, autonomous expert performance in practically any field. The review’s employment observations likewise do not establish that AI has had no labour-market effects; they say it found no visible break in the U.S. unemployment rate in the material examined.
Evidence to watch in future updates
Further updates to Thompson’s tracker may change individual statuses as new papers, filings and demonstrations appear. The most informative follow-up would be evidence that can be independently checked: published methods, reproducible results, clear accounting of human input and benchmarks that test performance across tasks rather than a single showcase problem.
For claims about broad autonomy, readers can also watch for systems completing multi-step physical tasks reliably outside controlled demonstrations, and for measured evidence of sustained effects on research work or employment. Until such evidence is available, the tracker is best read as one researcher’s framework for monitoring possible milestones, while the arrival of superintelligence remains unestablished by the review.
Key Questions
Does the review conclude that superintelligence has arrived?
No. It reports that Thompson’s tracker marks two indicators achieved, but says the tracker’s definitions and evidence do not establish that current AI systems are superintelligent.
What does a “partial” indicator mean?
Thompson says partial can include funded programs, research infrastructure, formal missions and other enabling work. It does not necessarily mean the system has nearly met the indicator.
How many claims did the review check?
The review examined 18 selected claims. 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 differ?
Thompson uses expert-level performance across practically any field. Other frameworks set different bars, including performance beyond large human organisations or highly autonomous systems that outperform humans at most economically valuable work.
Source: ThorstenMeyerAI.com
