🔍 Read the full analysis: 50 Signs Of Superintelligence, Checked: How Close Is AI To 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
A review published by ThorstenMeyerAI.com on October 26 checked 18 prominent claims from researcher Alan D. Thompson’s 50-sign superintelligence tracker. Seven were judged verified, five were vendor claims, three relied on a single source and three were disputed or contradicted; the tracker itself lists two signs as achieved, 33 as partial and 15 as not achieved. The assessment is not a formal measure of AGI or ASI, and the definitions of those terms vary.
A review published by ThorstenMeyerAI.com on October 26 found that evidence for prominent claims in researcher Alan D. Thompson’s 50-sign artificial superintelligence tracker is mixed: of 18 claims examined, seven were judged verified, five were vendor claims, three relied on a single source and three were disputed or contradicted. Thompson’s tracker lists two signs as achieved, 33 as partially achieved and 15 as not achieved, but those categories are his framework, not an independent measure of how close AI is to AGI or ASI.
The review describes Thompson, an Australian AI consultant who publishes The Memo newsletter, as maintaining a public checklist of 50 indicators ranging from mathematical results to speculative milestones. In his tracker, he calls Claude Mythos Preview and GPT-6 Astra “early ASI.” The review stresses that this is Thompson’s assessment, not a consensus finding or a conclusion established by the checked sources.
Thompson’s status labels also use a broad threshold. In an August correction note quoted by the review, he says “partially achieved” includes documented material progress, such as funded programs, research infrastructure, formal missions and other enabling work. That makes the 33 partial entries unsuitable to read as 33 developments nearly complete. The review gives two exoplanet candidates identified with Claude Code as an example of how enabling work can appear under a much larger milestone: the tracker places them under “first planet other than Earth colonized.”
The strongest verified examples in the review involve work that can be checked against primary records, including mathematical contributions and a result described as disproving a long-standing conjecture. The assessment is not uniformly dismissive: it reports that mathematicians Terence Tao, Don Knuth and Tim Gowers have described genuine AI contributions, while noting qualifications around coaching, existing literature and the kind of problem involved. Other headline claims, including some model-performance results, remain company-reported claims rather than independently reproduced findings.
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 Tracker’s Labels Matter
The tally matters because claims about “early ASI” can sound like a settled scientific status when they are based on a personal checklist with generous partial-credit rules. The review’s source-by-source approach gives readers a way to distinguish independently checked results from company announcements, single-source reporting and disputed claims. It also shows why a single count of signs cannot, by itself, establish that AI has reached a general capability threshold.
Some developments could still have practical importance without proving AGI or ASI. AI-assisted mathematical work, for example, may help researchers explore problems or produce formal proofs, even when humans guide the process or verify the result. By contrast, the review says there is no household-robot deployment at scale, no verifiably billion-dollar company run by AI, and no visible break in the U.S. unemployment rate attributable to AI. Those observations do not settle what AI systems can do in other settings, but they caution against treating technical demonstrations as evidence of broad economic autonomy.
For readers, the core issue is not whether a model can produce an impressive result once. It is whether it can do a wide range of valuable work reliably, with limited human direction, and whether that ability holds up under independent testing. Definitions, reproducibility and autonomy change the answer.
Products Worth Considering
Jackery Solar Generator 1000 V2 with 200W Solar Panel
LG UltraGear 32GX850A-B 32-Inch 4K UHD Glossy OLED Gaming Monitor
Jackery Explorer 300 Portable Power Station, 292Wh
As an affiliate, we earn on qualifying purchases.
Three Different AGI Yardsticks
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. The review compares that framing with Google DeepMind’s 2023 “Levels of AGI” scale, which distinguishes competent, expert, virtuoso and superhuman performance. Under that comparison, Thompson’s ASI roughly matches DeepMind’s “Expert” AGI level, rather than its superhuman category.
Google DeepMind’s June 2026 report, “From AGI to ASI,” uses a higher bar for superintelligence: 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. These definitions put different weight on breadth, performance and independence. A model that performs at an expert level on selected mathematical tasks could meet one person’s ASI description without meeting another organisation’s threshold for superintelligence.
The review also reports that Thompson’s separate “conservative countdown to AGI” stands at 99%. He says the remaining requirement is a robot that can carry out multi-step fine assembly, such as putting together furniture, at average-human reliability. That countdown is another subjective measure, not an agreed scientific estimate of the probability that AGI has arrived.
““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 quoted by ThorstenMeyerAI.com
What the Evidence Cannot Establish
The review’s verdicts apply to the 18 selected claims it checked, not all 50 tracker entries or every current AI capability. The source material does not provide a complete account of the evidence behind every verdict, and vendor claims may later be independently tested or revised. A “verified” result also does not establish that a system can perform reliably across fields or without substantial human input.
It remains unclear how much autonomy the cited mathematical systems demonstrated across the work as a whole. The review notes coaching, human feedback and proof-writing contributions in some cases, while describing other results as externally checked. It also does not establish that the absence of a visible unemployment-rate change means AI has had no labour-market effects; that measure alone cannot isolate AI’s influence. No shared threshold for AGI or ASI resolves these questions.
Evidence to Watch Next
The next useful signals will be independently reproducible results, clearer reporting on how much human direction a system needed, and demonstrations of reliable performance across tasks rather than isolated benchmarks. For the tracker’s remaining AGI countdown, Thompson identifies average-human reliability in multi-step fine assembly as the outstanding milestone. The source does not give a confirmed date for that capability or for any broader AGI or ASI determination.
Readers can also compare future tracker updates with primary research, company disclosures and external assessments, while checking whether a claimed result is a demonstration, a deployed capability or an independently validated outcome. Until definitions and evaluation methods are more consistent, claims that a model is “early ASI” should be treated as attributed judgments, not established status.
Key Questions
How many signs does Thompson’s tracker say have been achieved?
The tracker showed two achieved, 33 partially achieved and 15 not achieved as of October 26, 2026. Thompson’s partial category includes funded programs and enabling work.Does the review confirm that AI has reached ASI?
No. It checks selected claims and reports mixed results; it does not establish that AI has reached ASI. Thompson’s description of particular models as “early ASI” is his assessment.What did the 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.Why do definitions of ASI matter?
Thompson’s definition focuses on expert-level performance across fields, while Google DeepMind’s 2026 description sets a higher bar: capability beyond large organisations of humans. The same model may be labelled differently under each framework.What does Thompson identify as the remaining AGI milestone?
His separate countdown identifies a robot performing multi-step fine assembly at average-human reliability as the remaining milestone. This is his framework, not an agreed field-wide test.Source: ThorstenMeyerAI.com
