🔍 Read the full analysis: Is AI Near AGI Or ASI? Reviewing 50 Signs Of Superintelligence 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
A review of researcher Alan D. Thompson’s 50-indicator superintelligence tracker finds two indicators marked achieved, 33 partially achieved and 15 not achieved. Its author calls some leading models “early ASI,” but that label depends on a definition far less demanding than other prominent definitions of superintelligence.
A review published by Thorsten Meyer on October 26 examined 18 prominent claims from researcher Alan D. Thompson’s 50-sign superintelligence tracker, finding seven verified, five based on company claims, three reliant on a single source and three disputed or contradicted. Thompson’s tracker marks just two of its 50 indicators as achieved, while labeling 33 partial progress—a distinction that matters because his use of “ASI” is broader than some AI labs’ definitions of superintelligence.
Thompson, an Australian AI consultant who publishes The Memo newsletter, describes artificial general intelligence as a machine performing at the level of an average human and artificial superintelligence as a machine performing at expert-human level in practically any field. His public checklist ranges from mathematical discoveries and self-improving code to ambitious milestones such as colonizing another planet. In the tracker’s October 26 status grid, two indicators are achieved, 33 partially achieved and 15 not achieved.
The label “partially achieved” does not necessarily mean a capability is close to completion. In an August correction note, Thompson said it includes funded programs, research infrastructure, formal scientific missions and enabling work. Meyer’s review notes, for example, that two exoplanet candidates identified with Claude Code were listed under an indicator about colonizing a planet other than Earth. The tracker is therefore a collection of signals and judgments, not a direct measurement of how close AI is to superintelligence.
Among the strongest evidence Meyer identifies are mathematical contributions discussed by mathematicians Terence Tao, Don Knuth and Tim Gowers, along with a result on Erdős’s unit-distance conjecture supported by a companion paper from external mathematicians. The review also distinguishes public Lean certificates from a separate OpenAI account of results on decade-old problems, which it classifies as a vendor claim. For a claim to count as independently verified, the review sought evidence such as expert assessment or official documentation.
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 ASI Label Changes
The disagreement over definitions affects how readers interpret claims that current systems are “early ASI.” Thompson’s standard—expert-level performance in practically any field—is not the same as the bar used by Google DeepMind. Its 2023 framework grades capabilities from competent through expert and virtuoso to superhuman. On that scale, Thompson’s definition of ASI roughly corresponds to Expert AGI, not the superhuman category.
Google DeepMind’s June 2026 report uses a more demanding description of superintelligence: systems more capable than large organisations of humans. OpenAI’s charter, meanwhile, describes AGI in terms of highly autonomous systems outperforming people at most economically valuable work. These differences matter to the public debate: a model’s strong result in mathematics may be significant without showing broad autonomy or an ability to outperform human organisations across fields.
The review also separates narrow demonstrations from broader economic effects. It says no household robots are operating at scale, no AI-run company has been verifiably valued at a billion dollars, and there is no visible break in the U.S. unemployment rate attributable to AI. Those observations do not settle AI’s future impact, but they show why technical achievements alone cannot establish that society has reached superintelligence.
Products Worth Considering
Bose QuietComfort Ultra Earbuds (2nd Gen), Deep Plum
Apple iPad Air 13-inch (M4), 128GB, Wi-Fi 7, Space Gray
Samsung 27 Odyssey OLED G5 (G50SF) QHD QD-OLED Gaming Monitor
As an affiliate, we earn on qualifying purchases.
Three Standards for AI Progress
The tracker is one researcher’s framework, not a consensus benchmark. Its 50 entries cover disparate kinds of progress, making a single tally difficult to interpret: a mathematical proof, a funded research program and a physical-world milestone do not measure the same capability. Meyer presents the list as useful for identifying developments to watch, while cautioning that orange status means progress toward an indicator, not that the system is nearly there.
Thompson’s separate “conservative countdown to AGI” stands at 99%, according to the source material. He says the remaining step is a robot completing multi-step fine assembly, such as assembling flat-pack furniture, at average-human reliability. This is a separate measure from the 50-indicator tracker and should not be read as an independently established estimate of the probability or timing of AGI.
The source review also describes recursive self-improvement as real but limited. It cites Anthropic’s account that its best internal model scores 62.8% on a test for which the company says a full substitute for its researchers would need at least 85%. Anthropic also estimates that AI makes its research and development faster, but “not yet by a factor of 2.” These are company-provided assessments, not independent measurements.
“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 his tracker
What the Evidence Cannot Establish
The review’s 18-claim sample does not establish whether AI systems meet any single agreed definition of AGI or ASI. Its findings are a source check of selected claims, not a comprehensive audit of all 50 indicators. The source material does not provide the full results and citations for all 18 items, so the remaining claims cannot be evaluated individually here.
Several reported findings depend on company accounts, and even independently discussed mathematical results may involve human guidance. For example, the review says Claude found a construction for an open problem after coached exploration, with Don Knuth writing the proof; another reported result followed feedback and involved a problem that had been misformulated. How much work was autonomous varies by case and is not captured by a simple “solved” label.
It is also unclear how comparable the tracker’s indicators are or what evidence should count as achieving them. The “early ASI” description is Thompson’s view, not a finding accepted across the field. Neither the tracker’s percentages nor the review’s verdict counts establish when, or whether, systems will meet more demanding standards of broad capability and autonomy.
Further Checks on AI Milestones
The immediate next step for readers is to follow the primary evidence behind each claim: research papers, independent mathematical assessments, public filings and company disclosures. New results should be described with their conditions, including whether researchers coached the system, whether proofs are publicly checkable and whether other experts reproduced the work.
Thompson’s tracker is dated October 26, 2026, and the source review does not specify a scheduled update or an independent reassessment. Future versions may change the status of individual indicators as evidence develops. Until definitions and evaluation methods are more consistent, claims that AI is near AGI or ASI will remain dependent on which capability standard is being used and how independently the evidence has been checked.
Key Questions
Does the review conclude that AI has reached superintelligence?
No. It reports that Thompson marks two of 50 indicators achieved, but says his ASI definition is broader than definitions used by some AI labs. The review treats “early ASI” as Thompson’s characterization, not a field-wide conclusion.
What did the source review find among the 18 claims?
It classified seven as verified, five as vendor claims, three as based on a single source and three as disputed or contradicted. These figures describe the 18 selected claims, not all entries in the tracker.
What does a partial indicator mean?
Thompson says partial progress can include funded programs, research infrastructure, formal scientific missions and other enabling work. It does not necessarily mean a capability is nearly complete.
Why do definitions of ASI differ?
Thompson defines ASI around expert-level performance across practically any field. Google DeepMind’s 2026 description compares systems with large human organisations, while OpenAI’s AGI definition emphasizes highly autonomous performance across economically valuable work.
Source: ThorstenMeyerAI.com
