🔍 Read the full analysis: 50 Signs That Help Gauge AI’s Progress Toward AGI And ASI on ThorstenMeyerAI.com
Get business pricing on office and shipping supplies
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
A checklist by AI consultant Alan D. Thompson marks two of 50 proposed signs of artificial superintelligence as achieved, 33 as partly achieved and 15 as not achieved. A review of 18 prominent claims found seven verified, five based on company claims, three resting on a single source and three disputed or contradicted. The findings show notable advances in mathematics and research tools, but do not establish that superintelligence has arrived.
A review of 18 prominent claims in AI consultant Alan D. Thompson’s list of 50 signs of artificial superintelligence (ASI) found evidence for seven, while five rely on company statements and three are disputed or contradicted. Thompson’s tracker marks two signs achieved, 33 partly achieved and 15 not achieved, but those ratings do not establish that ASI has arrived.
Thompson, an Australian AI consultant who publishes The Memo newsletter, presents indicators ranging from mathematical discoveries and recursive code improvement to speculative milestones. His October 2026 video surveys entries recorded from January through October. In the tracker’s October 26 status grid, “partially achieved” includes funded programmes, research infrastructure and enabling work, not only systems close to meeting the stated goal.
The source article’s review checked 18 headline claims against papers, researcher accounts, company publications and official filings. It classified seven as verified, five as vendor claims, three as based on a single source, and three as disputed or contradicted. The review says the strongest independent support tends to come from results that can be examined directly, including mathematical work, a physics calculation and regulatory documents.
Examples include AI contributions to mathematical problems discussed by Terence Tao, Don Knuth and Tim Gowers. Those accounts describe meaningful results but also qualifications: human feedback or coaching, existing literature, and the particular style of problem all matter. A result being useful or surprising is not the same as showing general, autonomous expert ability 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
What the Evidence Can Establish
The checklist gives readers a way to track developments, but its categories combine milestones of very different kinds. A research programme or enabling technology may count as partial progress under Thompson’s rules even when the final capability remains absent. That makes the orange “partial” category a record of activity, not a countdown to a specific arrival date.
The review also highlights a practical distinction between demonstrated capability and claims about broader economic or societal effects. It reports no household robots operating at scale, no verifiably billion-dollar company run by AI, and no visible break in the U.S. unemployment rate. These observations do not settle what future systems may do, but they show why mathematical wins alone cannot answer whether AI has broadly displaced human work or become independently capable.
Definitions matter for policy, investment and public debate. Calling a model “early ASI” can imply a level of capability that other researchers reserve for systems surpassing large human organisations. Without a shared standard and repeatable tests, the same result can support very different conclusions.
Popular right now
De'Longhi Dinamica Plus Fully Automatic Espresso Machine
eufy E25 Robot Vacuum and Mop Combo with HydroJet Roller Mop
TCL 85 Designer Series QD-Mini LED TV (Model 85A400U) with Google Smart TV
As an affiliate, we earn on qualifying purchases.
Different Standards for Superintelligence
Thompson defines AGI as a machine performing at the level of an average human and ASI as a machine performing at an expert-human level in practically any field. By comparison, Google DeepMind’s 2023 “Levels of AGI” framework distinguishes competent, expert, virtuoso and superhuman performance. Under that scale, Thompson’s ASI description is closer to Expert AGI than to its superhuman category.
DeepMind’s June 2026 report uses a more demanding description of superintelligence: systems more capable than large organisations of humans. OpenAI’s charter defines AGI around highly autonomous systems that outperform humans at most economically valuable work. These benchmarks emphasize different qualities—expert performance, comparative capability and autonomy—so labels such as “proto-ASI” are difficult to interpret without specifying the standard.
Thompson’s separate conservative countdown to AGI is listed at 99%, with the final percentage tied to a robot completing multi-step fine assembly, such as assembling flat-pack furniture, at average-human reliability. That is one researcher’s chosen threshold, not a consensus measure of progress.
“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 the tracker
Where Capability Claims Still Fall Short
The review’s classifications do not amount to a comprehensive assessment of every item on the 50-sign list. The supplied source gives headline counts and examples, but not the full evidence and methodology for all 18 reviewed claims. It is also unclear how consistently the indicators are defined or whether the tracker’s status assignments have been independently audited.
Some results are described by their developers rather than reproduced independently. The source says Anthropic’s strongest internal model scored 62.8% on a test where the company estimated that replacing its researchers would require at least 85%. It also reports the company’s view that AI makes its research and development faster, but not by a factor of two. Those figures are company assessments; the source does not provide enough detail here to compare the test with other evaluations.
Thompson describes Claude Mythos Preview and GPT-6 Astra as “early ASI,” but that is his characterization, not an established field-wide conclusion. The evidence presented does not resolve whether current systems meet a consistent definition of AGI or ASI, how reliably they perform outside selected tasks, or how much human guidance contributed to particular results.
Evidence to Watch in Future Updates
The next useful test of the tracker will be whether its entries gain public, independently checkable evidence and whether the criteria for “partial” and “achieved” remain clear. Replicable results, published methods and evaluations that measure performance across different fields would help readers distinguish a demonstration from a broad capability.
Robotics is another milestone to watch in the source’s account. Its proposed final step in the conservative AGI countdown is reliable, multi-step fine assembly by a robot. The source also identifies the absence of household robots at scale and AI-run companies verifiably worth a billion dollars as gaps, while cautioning that these are not, by themselves, definitive tests of intelligence.
Further updates may change individual ratings as new papers, lab disclosures and independent checks appear. Until then, the 50 signs are best read as one researcher’s monitoring framework, not proof that a settled AGI or ASI threshold has been crossed.
Key Questions
Has the tracker confirmed that ASI has arrived?
No. Thompson’s tracker marks two of 50 indicators achieved, but those ratings do not establish that ASI has arrived. The source article treats the “early ASI” label as Thompson’s view, not a consensus finding.
What does “partially achieved” mean in the tracker?
Thompson says the category includes documented progress such as funded programmes, research infrastructure, formal scientific missions and enabling work. A partial rating does not necessarily mean the final capability is close.
How many of the reviewed claims were independently supported?
The source article classifies seven of 18 reviewed claims as verified. Five were vendor claims, three rested on a single source, and three were disputed or contradicted.
Why do definitions of ASI differ?
Thompson defines ASI as expert-human performance across practically any field. Other frameworks set different thresholds, including performance beyond humans or capabilities exceeding those of large human organisations.
Source: ThorstenMeyerAI.com
