🔍 Read the full analysis: 50 Signs Checked: Is AI Advancing Toward AGI And ASI? on ThorstenMeyerAI.com
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
An October 26 review of 18 prominent claims in Alan D. Thompson’s 50-indicator artificial superintelligence tracker classified seven as verified, five as vendor claims, three as based on a single source and three as disputed or contradicted. Thompson marks two of 50 indicators achieved, but his definition of ASI is narrower than some research labs’ definitions, and many tracker entries count funded or enabling work as partial progress.
Thorsten Meyer’s October 26 review of 18 prominent claims in researcher Alan D. Thompson’s 50-sign artificial superintelligence tracker found seven verified results, five claims resting on company reports, three based on a single source and three disputed or contradicted. The review highlights genuine advances in mathematics and other fields, but says the tracker’s status labels and differing definitions of superintelligence do not establish that AI has reached ASI.
Thompson, an Australian AI consultant who publishes The Memo newsletter, maintains a public checklist of 50 possible signs of ASI. Its entries range from mathematical results and recursive code improvement to speculative milestones. The status grid, last updated October 26, marks two indicators achieved, 33 partially achieved and 15 not achieved. Meyer describes the partial category as broad: Thompson’s August correction note says it includes documented progress such as funded programmes, research infrastructure, formal scientific missions and enabling work. It does not mean an indicator is nearly complete.
For his review, Meyer selected 18 strong claims from Thompson’s video and tracker and traced them to primary sources, including research papers, researcher accounts, company posts and official filings. Among the mathematics examples, Terence Tao described GPT-5.2’s contribution to an Erdős problem as achieved “more or less autonomously,” but noted it followed feedback and involved a problem that had been misformulated, with related results already known. Don Knuth reported that Claude found a construction during coached exploration; Knuth wrote the proof. Meyer also lists an OpenAI-reported disproof of a 1946 Erdős conjecture that was checked by external mathematicians.
Other entries need more qualification. The review labels a claim that GPT-6 Astra produced ten results on longstanding problems for about $2,000 a vendor claim, even though Lean certificates were made public; human preparation and verification remain relevant to interpreting the result. Meyer also points to Anthropic’s own account of limits to AI-assisted AI research: its best internal model scored 62.8% on a test that the company said would require at least 85% for a full replacement for its researchers. Anthropic estimated that AI made 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
What the Claims Establish
The review matters because claims that AI is nearing superintelligence can shape public expectations, business decisions and policy debates. A checklist of impressive demonstrations can help readers track developments, but a status label is not an independent measurement. The source check shows why each claim needs to be read with its evidence: a result verified by outside specialists carries a different weight from a company’s own benchmark or a project still at the funding stage.
The strongest confirmed examples in Meyer’s sample cluster where results can be inspected or checked, including mathematical work and formal records. That is evidence of meaningful capability in particular tasks, not proof that a system can reliably perform expert work across practically every field, operate autonomously or outperform large groups of people. The distinction affects how readers interpret both the tracker’s two achieved entries and the larger group marked partially achieved.
There are also practical tests that remain unmet, according to the source material. It reports no household robots operating at scale, no verifiable AI-run company worth a billion dollars and no visible break in the US unemployment rate. Those observations do not settle what AI systems can do, but they caution against treating research demonstrations as evidence that broad economic and physical-world changes have already arrived.
Products Worth Considering

Sony WH-1000XM6 Wireless Noise-Canceling Headphones, Black

Samsung 27-Inch Odyssey G5 QHD Gaming Monitor (G50F)

Bose QuietComfort Ultra Earbuds (2nd Gen), Wireless Noise-Cancelling Earbuds, Black
As an affiliate, we earn on qualifying purchases.
How the ASI Labels Differ
Thompson defines AGI as a machine performing at the level of an average, or median, human and ASI as a machine performing at expert-human level in practically any field. Those definitions are not shared universally. Google DeepMind’s 2023 “Levels of AGI” framework distinguishes competent, expert, virtuoso and superhuman performance. Under that scale, Thompson’s ASI definition roughly corresponds to Expert AGI, rather than the superhuman level.
DeepMind’s June 2026 report, “From AGI to ASI,” uses a more demanding description, defining superintelligence as a system 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. Autonomy, scope and the comparison group therefore change what a declaration of “early ASI” means. Thompson calls Claude Mythos Preview and GPT-6 Astra “early ASI,” but those are his assessments under his framework, not a shared scientific finding.
Thompson also runs a separate conservative AGI countdown that stood at 99% in the material Meyer reviewed. The remaining requirement, as Thompson describes it, is a robot completing multi-step fine assembly, such as putting together flat-pack furniture, with average-human reliability. That countdown is another indicator of Thompson’s view, not a consensus forecast or an independently validated probability.
“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 the tracker
What the Tracker Cannot Settle
The source check covers 18 selected claims, not all 50 indicators, and its findings do not establish the accuracy of every item in the checklist. The source material does not provide the full evidence and verdict for every claim, so the review’s overall count should not be read as a complete audit of the tracker. Some results can be checked independently; others rely on company disclosures, a single source or disputed interpretations.
There is no common definition of AGI or ASI in the accounts cited, and there is no single agreed test showing that a system performs at expert level across practically any field. The evidence also does not establish whether named systems can act reliably and autonomously in open-ended real-world settings. The gap between a mathematical achievement, a benchmark score and broad practical capability remains unresolved.
Several headline claims rest on definitions or classifications that readers may contest. For instance, a candidate exoplanet finding is listed under an indicator about colonising a planet other than Earth, while the tracker’s broad partial-progress rule includes enabling work. That classification is not evidence that colonisation has occurred. The review also does not provide enough information to determine the real-world economic effect of AI from unemployment figures alone.
Evidence Needed for Broader Claims
The next useful test is whether claims can be reproduced and independently examined, with clear information about human guidance, tools, costs and the work required to verify results. For mathematics, that means distinguishing a model’s proposed solution from the proof and checking what researchers contributed. For company benchmarks, readers need the test conditions and independent replication before treating reported scores as established performance.
Progress on physical tasks will also be relevant to Thompson’s stated AGI countdown: a robot would need to complete multi-step fine assembly at average-human reliability. The source material gives no date for that milestone. Until results on such tasks, AI-led research and broader autonomous work are demonstrated and evaluated consistently, the 50-sign list is best understood as one researcher’s monitoring framework, not a definitive measure that ASI has arrived.
Key Questions
How many indicators does Thompson’s tracker mark as achieved?
Its status grid, last updated October 26, 2026, marks two of 50 indicators achieved, 33 partially achieved and 15 not achieved.
What did Meyer find when he checked 18 claims?
Meyer classified seven as verified, five as vendor claims, three as based on a single source and three as disputed or contradicted. The review selected 18 claims and does not amount to an audit of every tracker entry.
Does the tracker show that ASI has arrived?
No. It records Thompson’s assessment using his own definition and broad partial-progress category. Meyer’s review identifies verified advances but does not establish that AI systems meet a shared definition of ASI.
Why do definitions of ASI matter?
Thompson’s definition refers to expert-human performance in practically any field. Google DeepMind’s cited framework distinguishes expert performance from superhuman performance, while its 2026 report describes superintelligence in relation to large organisations of humans. Those standards can lead to different conclusions about the same model.
What milestone does Thompson say remains in his AGI countdown?
Thompson says the final step is a robot completing multi-step fine assembly, such as assembling flat-pack furniture, at average-human reliability. The source material does not give a timetable for achieving it.
Source: ThorstenMeyerAI.com
