🔍 Read the full analysis: 50 Superintelligence Signals: How Close Is AI To AGI And ASI? 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 published by ThorstenMeyerAI.com on October 26 examines claims in researcher Alan D. Thompson’s 50-indicator tracker for artificial superintelligence. Thompson marks two indicators achieved, 33 partially achieved and 15 not achieved; a separate check of 18 headline claims found seven verified, while others were vendor claims, based on one source, disputed or contradicted.
ThorstenMeyerAI.com published a review on October 26 of 18 prominent claims in AI consultant Alan D. Thompson’s list of 50 indicators of artificial superintelligence, finding seven verified, five supported only by vendor claims, three resting on a single source and three disputed or contradicted. Thompson’s tracker labels two indicators achieved, 33 partially achieved and 15 not achieved—a tally that offers a snapshot of progress, not proof that superintelligence has arrived.
The review traces selected claims from Thompson’s October video and public checklist to papers, researchers’ accounts, laboratory posts and official filings. Its strongest verified examples are concentrated in areas where results can be inspected, including mathematical work, a physics calculation and regulatory records. The review says that verification does not mean every dramatic summary of those results is accurate: some mathematical contributions involved coaching, particular problem formulations or substantial human work.
Among the examples, mathematician Terence Tao described an AI contribution to an Erdős problem as more or less autonomous after feedback, while noting that the problem had been misformulated and that similar results existed. Donald Knuth documented a construction produced during coached exploration with Claude; Knuth wrote the proof, and later work addressed another case. The review also identifies an OpenAI claim about disproving a 1946 unit-distance conjecture as externally verified by mathematicians. By contrast, claims about results from GPT-6 Astra are classified as vendor claims in the review, even where Lean certificates are public, because the reported results and costs have not been independently reproduced there.
Thompson’s categories need careful reading. His August correction says “partially achieved” covers documented progress that can include funded programmes, research infrastructure and enabling work. It does not mean a system has nearly met the indicator. For instance, two exoplanet candidates reportedly identified with Claude Code are listed against an indicator about colonizing another planet, a connection the review presents as a broad interpretation rather than evidence of colonization.
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 scorecard needs scrutiny
The tracker draws attention to concrete developments, but its totals can be misleading if readers treat every orange, or partially achieved, entry as evidence that AI is close to a specific threshold. The review’s audit shows how much depends on what counts as progress and how claims are checked. A mathematical result that can be independently examined is not equivalent to a laboratory’s own benchmark, and neither establishes broad, autonomous competence across fields.
Definitions also change the answer to the question in the headline. Thompson defines artificial general intelligence as performance at the level of an average human and artificial superintelligence as performance at expert-human level in practically any field. Google DeepMind’s 2023 framework calls its 90th-percentile category “Expert” and distinguishes it from “Superhuman.” Its June 2026 report describes superintelligence as exceeding the cognitive capabilities of large human organisations. OpenAI’s charter, meanwhile, emphasizes highly autonomous systems that outperform people at most economically valuable work. These are different tests, so a system could meet one person’s threshold without meeting another’s.
The distinction matters to readers weighing claims about employment, research and the pace of AI development. The review says there is no visible break in the US unemployment rate attributable to AI and no verified AI-run company worth a billion dollars. It also reports that household robots have not reached scale. Those absences do not settle what future systems may do, but they caution against treating research demonstrations as evidence that broad economic or physical-world transformation has already happened.
Popular right now
Philips 4400 Series Fully Automatic Espresso Machine with LatteGo Milk System
De'Longhi Stilosa Compact Manual Espresso Machine with Milk Frother
Dyson V8 Cordless Vacuum
As an affiliate, we earn on qualifying purchases.
How the indicators are defined
Thompson, an Australian AI consultant who publishes The Memo newsletter, maintains a public checklist spanning technical and physical-world signs, from recursive code improvement to much more speculative milestones. The October montage covers entries from January through October 2026. The tracker’s dated status grid records 2 achieved, 33 partially achieved and 15 not achieved; the review advises reading it as one researcher’s framework rather than a standardized measurement.
Thompson also maintains a separate “conservative countdown to AGI” that he puts at 99%. In his account, the remaining step is a robot able to complete multi-step fine assembly, such as assembling flat-pack furniture, with average-human reliability. That estimate is tied to his chosen test, not a generally accepted probability that AGI is imminent. DeepMind and OpenAI use different definitions and benchmarks, while the review’s claim-by-claim audit applies its own evidence labels. The tracker, countdown and audit measure different things and should not be combined into a single consensus score.
“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 to his tracker
Evidence and definitions still differ
The source review covers 18 selected claims, not all 50 indicators, and its findings cannot be treated as a complete audit of the tracker. The supplied account does not give the full list of claims in the five vendor-only, three single-source and three disputed or contradicted categories, so those results cannot be assessed individually here. Public certificates or company descriptions may help scrutiny, but the review distinguishes them from independent reproduction.
It is also unclear how the tracker’s status labels are applied consistently across unlike milestones, or what evidence would move each partially achieved item to achieved. The disagreement over AGI and ASI definitions means there is no shared threshold against which to calculate a definitive distance to either. Finally, the review’s observations about employment, robots and company valuations are time-bound; they do not establish what will happen in later periods.
Independent checks will shape the picture
The immediate next step is further scrutiny of the individual claims: whether researchers outside the labs can reproduce reported results, whether papers and formal proofs support the summaries, and whether systems can carry out tasks with less coaching. For claims tied to deployment, readers will need evidence of sustained performance in real settings rather than a demonstration or funded programme. Those tests are more informative than a headline tracker total.
Thompson’s checklist is dated October 26, 2026, and the review does not identify a scheduled update or an agreed external audit. The status of its entries may change as further evidence appears. Until then, the defensible conclusion is limited: notable AI contributions are documented in some fields, but the reviewed claims do not establish that systems meet a broadly shared definition of AGI or ASI.
Key Questions
Does the tracker say artificial superintelligence has arrived?
No. It marks two of 50 indicators achieved, 33 partially achieved and 15 not achieved. Those statuses are Thompson’s framework, not a consensus finding that ASI exists.
What did the review find when it checked 18 claims?
It 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 did not audit every indicator.
Why do people disagree about whether AI is near AGI or ASI?
They use different definitions. Thompson’s ASI threshold is expert-human performance in practically any field, while DeepMind describes superintelligence in relation to the capabilities of large human organisations. OpenAI’s charter emphasizes autonomy and economically valuable work.
Does “partially achieved” mean a milestone is almost complete?
Not necessarily. Thompson says the label can include funded programmes, research infrastructure and enabling work. It records progress under his criteria, not a measured percentage of completion.
What remains unproven in the review?
The review does not establish that AI systems meet a shared definition of AGI or ASI. It also reports no household robots at scale, no verifiable billion-dollar AI-run company and no visible break in US unemployment attributable to AI, while noting that its audit covered only selected claims.
Source: ThorstenMeyerAI.com
