🔍 Read the full analysis: 50 Indicators Of Superintelligence And The Distance 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 October 26 examined 18 claims featured in Dr. Alan D. Thompson’s 50-indicator artificial superintelligence tracker. It classified seven as verified, five as vendor claims, three as relying on a single source and three as disputed or contradicted; the tracker itself lists two indicators achieved, 33 partially achieved and 15 not achieved.
A review published on October 26, 2026, checked 18 prominent claims from researcher Alan D. Thompson’s 50-indicator artificial superintelligence tracker against papers, lab statements and other primary sources. The review classified seven claims as verified, five as vendor claims, three as based on a single source and three as disputed or contradicted, while stressing that the tracker’s definition of ASI differs from definitions used by major AI labs.
Thompson’s public tracker showed two of 50 indicators achieved, 33 partially achieved and 15 not achieved as of October 26. The review says the “partially achieved” category covers a broad range of activity, including funded programmes, research infrastructure and enabling work. Thompson’s August correction note describes orange status as “documented material progress toward an indicator”; it does not mean a milestone is nearly complete.
The review found the strongest substantiation among claims that could be checked against mathematical work or formal records. It cites contributions to mathematical problems discussed by Terence Tao and Don Knuth, as well as an OpenAI-associated result on a long-standing conjecture that was examined by external mathematicians. Those examples come with qualifications: some work involved prompting or coaching, and in one case a researcher wrote the proof.
Other claims remain tied to company reporting or narrower evidence. The review labels a claim about GPT-6 Astra producing ten formalized results as a vendor claim, noting that Lean certificates were public but humans prepared them. On recursive AI research, it cites Anthropic’s assessment that its best internal model scored 62.8% on a test, below the company’s stated 85% threshold for replacing its researchers, and that AI had made research and development faster but not by a factor of two.
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 ASI Claims Need Clear Benchmarks
The review matters because labels such as “early ASI” can suggest a level of capability that different researchers and labs do not define in the same way. A tracker can be useful as a watchlist, but its status categories and thresholds shape what readers infer from the same evidence. Distinguishing completed results from funded work, company-reported performance and independently checked findings helps prevent those categories from being read as equivalent.
The evidence also points to an uneven picture of progress. The review identifies credible AI contributions in areas such as mathematics and formal verification, where outputs can be examined. It does not establish broad autonomous competence across fields, nor does it show that systems can independently run research organizations or perform complex physical tasks at human reliability. That gap matters to readers weighing claims about economic effects, safety and how close systems are to general-purpose autonomy.
Products Worth Considering
Amazon Kindle Paperwhite (12th Generation, 2024), 16 GB, 7-inch Display, Jade
Apple iPad Air (2025) 13‑inch Wi‑Fi 256 GB – Space Gray (M3 chip)
Apple AirPods Pro 3 Wireless Earbuds with Active Noise Cancellation
As an affiliate, we earn on qualifying purchases.
Different Definitions, Different Milestones
Thompson defines AGI as a machine performing at the level of an average human and ASI as performing at the level of an expert human in practically any field. Google DeepMind’s 2023 framework instead separates levels including competent, expert, virtuoso and superhuman performance. Under that scale, expert performance is not the same as superhuman performance.
Google DeepMind’s June 2026 report describes superintelligence as exceeding the cognitive capabilities of large organizations of humans. OpenAI’s charter defines AGI around highly autonomous systems that outperform people at most economically valuable work. The review argues that these distinctions affect how claims in Thompson’s tracker should be interpreted: a system may perform strongly in a particular expert domain without meeting a broader definition of ASI.
The tracker also sits alongside a separate countdown by Thompson that stood at 99% toward AGI. According to the source review, its remaining requirement is a robot capable of multi-step fine assembly, such as assembling furniture, at average-human reliability. That figure is Thompson’s assessment, not a standardized industry measure.
“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
What the Tracker Cannot Establish
The review is a check of 18 selected claims, not a comprehensive audit of all 50 indicators or all current AI capabilities. Its classifications also depend on the evidence available and the review’s standards for verification. Company claims that have not been independently reproduced may change as further documentation or testing becomes available.
It remains unclear whether the systems described as “early ASI” meet any common, independently agreed threshold for artificial superintelligence. The tracker’s partial category includes work at very different stages, while the definitions used by Thompson and AI labs are not interchangeable. The source material also says there is no visible break in the US unemployment rate, no household robots deployed at scale and no verifiably billion-dollar AI-run company; it does not provide data or a method for independently testing each of those broader observations.
Evidence to Watch in Future Updates
Further tracker updates and primary-source documentation will show whether partially achieved indicators move to achieved status, remain incomplete or are revised. The clearest evidence would include independently reproducible results, transparent evaluation methods and demonstrations of sustained performance beyond a narrow task.
For claims about autonomy, the tracker’s own stated benchmark makes reliable multi-step physical assembly one marker to watch. For research acceleration, Anthropic’s published evaluations and future reporting may clarify whether AI tools can take on more of the work performed by researchers. Until comparable evidence is available, the review’s findings support treating “early ASI” as Thompson’s characterization rather than an established industry-wide classification.
Key Questions
How many indicators does Thompson’s tracker mark as achieved?
As of October 26, 2026, it marked two of 50 achieved, 33 partially achieved and 15 not achieved.
What did the review find in its sample of 18 claims?
It classified seven as verified, five as vendor claims, three as relying on a single source and three as disputed or contradicted.
Does Thompson’s definition of ASI match Google DeepMind’s?
No. Thompson describes ASI as expert-human performance in practically any field. Google DeepMind’s 2026 report describes superintelligence as exceeding the cognitive capabilities of large organizations of humans.
Does the tracker show that artificial superintelligence has arrived?
No independent consensus or shared benchmark is established by the tracker. The review says Thompson’s “early ASI” label reflects his own framework, and that evidence for the selected claims varies in strength.
Source: ThorstenMeyerAI.com
