🔍 Read the full analysis: AI And Superintelligence: 50 Signs Checked For Clues To AGI And ASI on ThorstenMeyerAI.com
Get hardware and tech essentials 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
A review published by ThorstenMeyerAI.com on October 26 examined 18 claims drawn from AI researcher Alan D. Thompson’s 50-indicator artificial superintelligence tracker. It classified seven as verified, five as vendor claims, three as based on a single source and three as disputed or contradicted; the tracker itself lists two indicators achieved, 33 partially achieved and 15 not achieved.
ThorstenMeyerAI.com published a review on October 26 checking 18 claims featured in AI consultant Alan D. Thompson’s 50-indicator artificial superintelligence tracker. The review classified seven claims as verified, five as vendor claims, three as single-source claims and three as disputed or contradicted, underscoring how uneven the evidence is behind assertions that advanced AI systems are nearing superintelligence.
Thompson’s public tracker listed two of its 50 indicators as achieved, 33 as partially achieved and 15 as not achieved when it was last updated October 26, according to the review. The categories need qualification: Thompson says “partially achieved” includes funded programmes, research infrastructure, scientific missions and other enabling work. It does not mean an indicator is nearly complete or that the stated outcome has occurred.
The review says the strongest independently checkable examples cluster in mathematics and other areas with public records. It describes AI contributions to mathematical problems associated with Terence Tao and Don Knuth as genuine, while noting human coaching, proof-writing and limits tied to the problems themselves. It also reports that an OpenAI model’s claimed disproof of a longstanding conjecture was supported by work from external mathematicians. These examples show useful research contributions, not a general ability to perform expert work autonomously in every field.
Other prominent claims have weaker support. The review labels results attributed to companies as vendor claims when the reported figures have not been independently reproduced, even if materials such as formal proof certificates are public. It also highlights gaps between AI-assisted research and full self-replacement: citing Anthropic’s own assessment, the article says its best internal model scored 62.8% on a test for which the company estimated a full replacement for its researchers would need at least 85%. Anthropic also estimated its research and development work was faster with AI, 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
Evidence Falls Short of an ASI Verdict
The review matters because “ASI” does not have one agreed threshold, and a checklist can make progress appear more definite than it is. Thompson defines artificial superintelligence as a machine performing at expert-human level in practically any field. The article says that definition is closer to Google DeepMind’s “Expert AGI” category than to DeepMind’s 2026 description of superintelligence as exceeding the cognitive capability of large organisations of humans.
That distinction changes how readers should interpret claims such as “early ASI.” A system may make a valuable contribution to a mathematical problem without being broadly capable, reliably autonomous or superior to human organisations. Verification, scope and independence of evidence matter: an independently checked result carries a different weight from a company’s own benchmark or an indicator that counts preliminary work.
The review also points to practical milestones that remain absent in its account, including household robots operating at scale, a verifiably billion-dollar company run by AI and a visible break in the US unemployment rate attributable to AI. Their absence does not settle when or whether such outcomes will occur, but it cautions against treating selected research results as evidence that broad economic or social transformation has already arrived.
Products Worth Considering
Apple MacBook Air 15-inch Laptop with M5 Chip, 16GB Memory, 512GB SSD, and Liquid Retina Display
LG 34GX90SA-W 34-inch Ultragear WQHD OLED Curved Gaming Monitor
LG UltraGear 45GX900A-B 45-Inch Curved OLED Gaming Monitor
As an affiliate, we earn on qualifying purchases.
Different Definitions, Different Thresholds
Thompson is an Australian AI consultant who publishes The Memo newsletter and maintains the public checklist. His tracker ranges from recursive code self-optimization to much more speculative indicators. The source review describes the tracker as one researcher’s framework, rather than a standard scientific measurement or a consensus assessment of AI capability.
Definitions help explain the disagreement. Google DeepMind’s 2023 “Levels of AGI” framework distinguishes “Competent,” “Expert,” “Virtuoso” and “Superhuman” performance. The source says Thompson’s expert-level definition of ASI roughly corresponds to Expert AGI on that scale. DeepMind’s 2026 report uses a higher bar, describing superintelligence in relation to the capabilities of large human organisations. OpenAI’s charter, meanwhile, defines AGI around highly autonomous systems that outperform people at most economically valuable work.
The review’s approach was to select 18 of the strongest claims from Thompson’s video and tracker, then trace them to papers, lab posts, researchers or official filings and look for critics. That method can test whether particular claims have a solid source. It cannot, by itself, determine whether a broad system meets every possible definition of AGI or ASI.
““documented material progress toward an indicator, including funded programs, research infrastructure, formal scientific missions, and enabling work””
— Alan D. Thompson, in a correction note cited by ThorstenMeyerAI.com
What the Tracker Cannot Establish
The review’s classifications are not a verdict on every AI capability or on the timing of AGI or ASI. The source material does not provide the full evidence and reasoning for all 18 individual assessments, and some claims remain vendor-reported, based on one source or disputed. Public proof materials can support a mathematical result without establishing that a system produced it independently or can repeat the performance across unrelated tasks.
It is also unclear how much weight to give the tracker’s 33 partial indicators, because its definition includes enabling work rather than only demonstrated outcomes. Nor does the review establish that AI caused or did not cause changes in employment, company valuations or robotics adoption. Those questions require additional evidence and clearly defined measures. The term “early ASI” remains a claim tied to Thompson’s definitions, not a generally accepted finding.
Further Checks Needed on Capability Claims
The next useful developments will be independent replication, clearer reporting of methods and evidence of reliable performance across domains. For claims involving mathematical results, researchers can examine the formal proofs and document the degree of human guidance. For company-reported benchmarks, outside testing would help show whether results hold beyond internal evaluations.
Readers can also track whether the indicators move from funded or preparatory work to demonstrated outcomes, while keeping the definitions attached to each claim. The source review does not announce a formal follow-up date or a new independent assessment. Until more evidence is available, the tracker is best read as a list of proposed signals to watch, not confirmation that superintelligence has arrived.
Key Questions
What did the review find about the 18 ASI claims?
ThorstenMeyerAI.com classified seven as verified, five as vendor claims, three as single-source claims and three as disputed or contradicted. The review says it traced the claims to primary sources and sought criticism, though the source material supplied here does not detail every verdict.
How many indicators does Thompson say have been achieved?
The tracker listed two achieved, 33 partially achieved and 15 not achieved as of October 26, 2026. Thompson’s “partially achieved” category includes enabling work and funded programmes, so it does not mean those indicators are close to completion.
Does the review confirm that AI superintelligence has arrived?
No. The review checks specific claims and reports some verified AI research contributions, but it does not establish that a system meets a broadly accepted definition of artificial superintelligence. Definitions differ, and several claims remain unverified or disputed.
Why do definitions of ASI matter?
Thompson’s definition, as described in the review, is expert-human performance in practically any field. Other frameworks set different thresholds, including performance beyond large organisations of humans. A claim that a model is “early ASI” can mean different things depending on which benchmark and definition are being used.
Source: ThorstenMeyerAI.com
