🔍 Read the full analysis: What 50 Signs Reveal About AI’s Progress Toward Superintelligence on ThorstenMeyerAI.com
Get the latest gadgets 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
Thorsten Meyer’s October 26 review checked 18 claims from researcher Alan D. Thompson’s 50-indicator AI 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 as achieved and 33 as partially achieved. The review finds credible progress in mathematics and AI-assisted research, but says definitions and generous partial-credit criteria make the tracker a poor measure of whether superintelligence has arrived.
A review published by ThorstenMeyerAI.com on October 26 examined 18 claims in researcher Alan D. Thompson’s 50-indicator AI tracker, finding seven verified, five based on company claims, three supported by a single source and three disputed or contradicted. The review challenges the idea that those indicators show artificial superintelligence has arrived: Thompson’s tracker marks just two of 50 indicators achieved, while its definition of superintelligence is less demanding than some major AI labs’ definitions.
Thompson, an Australian AI consultant who publishes The Memo newsletter, maintains a public checklist covering indicators that range from mathematical discoveries and automated research to physical-world capabilities. In an October video, he described models including Claude Mythos Preview and GPT-6 Astra as “early ASI.” Those are his characterizations, not a consensus finding or an independently established classification.
The tracker’s October 26 status grid records two indicators as achieved, 33 as partially achieved and 15 as not achieved. In an August correction note, Thompson said “partially achieved” can include funded programs, research infrastructure, formal scientific missions and other enabling work. That means the category does not necessarily indicate that a system can perform the end task. The review gives as an example two exoplanet candidates reportedly found with Claude Code, listed under an indicator about colonizing another planet.
Among the reviewed claims, the strongest evidence was in mathematics, where the review traced results to researchers’ accounts, papers or other primary material. It reports that Terence Tao described GPT-5.2 as solving an Erdős problem “more or less autonomously,” but with feedback; the problem had been misformulated and similar results existed. Don Knuth’s account of a Claude-assisted construction also involved coached exploration, and Knuth wrote the proof. Other cases included an OpenAI model’s work on a unit-distance conjecture, which the review says was checked by external mathematicians. By contrast, several claims about model performance and research output relied on company announcements rather than independent reproduction.
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 review’s central finding is not that AI progress is insignificant, but that progress on specific tasks does not by itself establish superintelligence. A mathematical proof, a research lead or a model’s high score can show useful capability. It does not automatically show broad, reliable performance across fields, independent operation or the ability to outperform large groups of people.
That distinction matters because claims about superintelligence can shape public expectations, investment and policy discussions. If “partially achieved” includes preparatory programs and infrastructure, the tracker can signal activity without showing the capability itself. Readers comparing its status counts with other measures need to know what each label means and whether the underlying result was independently checked.
The review also points to evidence against a simple story of runaway self-improvement. It cites Anthropic’s assessment that its best internal model scored 62.8% on a test the company says would require at least 85% for a full replacement for its researchers. Anthropic estimated that AI had made its research and development faster, “but not yet by a factor of 2.” These are company-reported assessments, not independent measurements, but they complicate claims that AI systems already replace the teams building them.
Products Worth Considering
Sony WH-1000XM6 Wireless Noise-Canceling Over-Ear Headphones, Blue, Bundle
Jackery Explorer 300 Portable Power Station, 292Wh
Logitech MX Keys S Wireless Keyboard, Graphite
As an affiliate, we earn on qualifying purchases.
Different Definitions, Different Thresholds
Thompson defines artificial general intelligence as a machine performing at the level of an average human, and artificial superintelligence as one performing at the level of an expert human in practically any field. The review argues that this use of “ASI” differs from other published frameworks and can make the label appear closer than it is under stricter definitions.
Google DeepMind’s 2023 “Levels of AGI” framework distinguishes levels such as competent, expert, virtuoso and superhuman. The review says Thompson’s ASI definition roughly corresponds to expert-level AGI in that framework, rather than its superhuman level. A June 2026 DeepMind report describes superintelligence as capability beyond that of large human organizations. OpenAI’s charter, meanwhile, defines AGI around highly autonomous systems that outperform humans at most economically valuable work.
The tracker is therefore best read as one researcher’s selection of developments to watch, not as a standardized scientific test. Meyer’s review says it examined a selected set of 18 prominent claims, not all 50 indicators, and checked them against primary material where available. Its findings provide a snapshot of the evidence and its limitations as of the publication date.
“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 classifications are not a formal audit of every tracker entry. It selected 18 claims, and the supplied account does not give the full claim-by-claim table or enough detail to independently evaluate all 18 verdicts. The seven “verified” findings also cover particular claims, not a general conclusion that the systems involved meet any agreed definition of superintelligence.
Several claims remain dependent on vendors’ own reports or limited sourcing. Public Lean certificates, for example, may allow parts of a mathematical result to be checked, while the review still classifies a reported GPT-6 Astra result as a vendor claim because people prepared the problems. The review also says no household robots are operating at scale, it found no verifiably billion-dollar AI-run company, and it saw no visible break in the U.S. unemployment rate. Those observations do not settle what future systems may do or establish that AI has had no labor-market effects.
It is also unclear how consistently different reviewers would apply labels such as “verified” or “partially achieved,” and whether the tracker’s indicator set can measure a concept as broad as superintelligence. Thompson’s claim that some models are “early ASI” remains his interpretation; the review does not establish that the claim is accepted by independent researchers or AI labs.
Evidence to Watch in Future Updates
The next useful test will be whether future tracker updates connect each status change to reproducible results, clearly stated definitions and evidence that can be checked outside the company or researcher making the claim. Readers can compare new mathematical results with published proofs and expert assessments, and distinguish demonstrated capabilities from funded work or announced research plans.
Thompson separately runs a countdown to AGI that he places at 99%. The review says he reserves the final percentage for a robot able to complete multi-step fine assembly, such as assembling flat-pack furniture, with average-human reliability. That is a milestone in his framework, not a universally agreed threshold. No timetable for reaching it is established in the material reviewed.
For now, the October review supports a measured conclusion: AI systems have made documented contributions in some areas, particularly mathematics, while major claims of broad autonomy and superintelligence remain unevenly supported. Further independent evaluations and clearer agreement on what “ASI” means would help establish whether that assessment changes.
Key Questions
What did the review find about the 50-sign tracker?
It checked 18 selected claims: seven were classified as verified, five as vendor claims, three as based on a single source and three as disputed or contradicted. The tracker itself listed two indicators achieved, 33 partially achieved and 15 not achieved on October 26, 2026.
Does the tracker show that superintelligence has arrived?
No. The review treats the tracker as one researcher’s framework, not a standardized measurement. It says several claims show real progress, but the definitions and evidence do not establish that AI systems meet a widely accepted standard for superintelligence.
Why does the meaning of “partially achieved” matter?
Thompson says the label can cover funded programs, research infrastructure and enabling work. A partial status can therefore reflect activity toward a goal without showing that an AI system can perform the capability itself.
What evidence was strongest in the review?
The review found the clearest support in mathematical results that could be traced to researchers, papers or external checks. It also describes important qualifications, including human feedback, coached exploration and problem preparation.
What remains uncertain about AI’s progress?
The review examined only a selection of claims, and several depended on vendor reports or limited sourcing. There is no agreed definition of superintelligence in the material, and the evidence does not establish when—or whether—systems will meet the capabilities described in the tracker.
Source: ThorstenMeyerAI.com
