🔍 Read the full analysis: Understanding AI’s Progress: 50 Signs Of Superintelligence, AGI, And ASI 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
A review published by ThorstenMeyerAI.com on October 26 examined claims in researcher Alan D. Thompson’s 50-sign ASI tracker. It found substantial, independently checkable progress in mathematics and other fields, but the tracker’s broad “partially achieved” category and varying definitions do not establish that current models are superintelligent.
ThorstenMeyerAI.com published a review on October 26 of claims in AI consultant Alan D. Thompson’s 50-indicator artificial superintelligence tracker, reporting that the evidence ranges from independently checked mathematical results to company claims and disputed assertions. The review says Thompson marks two indicators achieved, 33 partially achieved and 15 not achieved, but warns that the “partially” category includes research programmes and enabling work, not just systems nearing the stated goal.
The review’s author says they traced 18 prominent claims from Thompson’s 2026 video and tracker to papers, lab posts, researcher notes or regulatory filings. Their assessment classified seven as verified, five as vendor claims, three as resting on a single source, and three as disputed or contradicted. These are the reviewer’s classifications, not a formal audit or an agreed scientific standard.
Among the examples described as verified are AI contributions to mathematical problems associated with Terence Tao and Don Knuth, as well as a conjecture result attributed to an OpenAI model and supported by a paper involving external mathematicians. The account also includes caveats: some work involved human feedback or coaching, and one problem had been misformulated or had related results already. For an OpenAI claim about decade-old problems solved and formalised in Lean, the review uses the label vendor claim, while noting that Lean certificates are public.
The article also highlights limits to claims of AI-driven self-improvement. It reports that Anthropic’s 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 said its research and development work was faster with AI, “but not yet by a factor of 2.” These figures and assessments are attributed to the company; they do not independently establish how AI affects research productivity across the industry.
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 Behind the ASI Labels
The review matters because a checklist can make very different kinds of evidence appear comparable. A demonstrated mathematical result, a funded programme, a research tool and a speculative milestone are not equivalent accomplishments. Thompson’s tracker counts 33 indicators as partially achieved, but its definition, as quoted in the review, includes “documented material progress” such as infrastructure, formal missions and enabling work. That status signals activity toward a goal, not proof that the goal is close or accomplished.
Definitions also change the interpretation. Thompson defines ASI as a machine performing at expert-human level in practically any field. The review compares this with Google DeepMind’s 2023 capability scale, where “Expert” is one level below “Virtuoso” and “Superhuman.” DeepMind’s June 2026 report describes superintelligence as exceeding the cognitive capability of large human organisations. Calling a model “early ASI” therefore depends on which definition is being used; the label is not a settled measurement shared by the field.
For readers, the distinction affects how to weigh claims about jobs, scientific discovery and the pace of AI development. The review says it found no visible break in the US unemployment rate, no household robots operating at scale, and no verifiably billion-dollar AI-run company among the signs it considered. Those observations do not settle the long-term effects of AI, but they caution against treating a list of possible indicators as evidence that broad economic or physical-world transformation has already occurred.
Products Worth Considering
Jackery HomePower 3600 Plus Portable Power Station, 3584Wh, 3600W
Logitech MX Master 4 for Mac, Ergonomic Bluetooth Mouse with Haptic Feedback, Ultra-Fast Scrolling, USB-C Charging, for MacOS, iPadOS, Windo
Samsung 27-inch Odyssey OLED G6 QD-OLED QHD Gaming Monitor
As an affiliate, we earn on qualifying purchases.
How Thompson Defines Progress
Thompson, an Australian AI consultant who publishes The Memo newsletter, maintains a public list of 50 indicators ranging from mathematical results to speculative milestones. In the tracker’s status grid, last updated October 26, he describes Claude Mythos Preview and GPT-6 Astra as “early ASI.” That wording is Thompson’s assessment. The source article does not establish that the companies behind those models or the wider research community share it.
The tracker’s orange, or “partially achieved,” label has a broad meaning. In an August correction note quoted by the review, Thompson said it can cover funded programmes, research infrastructure, formal scientific missions and other enabling work. The article gives the example of two exoplanet candidates reportedly identified with Claude Code being placed under an indicator about the first planet other than Earth being colonized. That classification illustrates the distance that can remain between evidence relevant to a topic and achievement of the indicator as written.
Other organisations use different benchmarks. Google DeepMind’s 2023 framework grades abilities from “Competent” through “Expert,” “Virtuoso” and “Superhuman.” OpenAI’s charter describes AGI in terms of highly autonomous systems that outperform humans at most economically valuable work. Capability, autonomy and breadth are distinct measures, so one result in mathematics or coding cannot by itself show that a system meets every 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 an August correction note quoted by ThorstenMeyerAI.com
What the Tracker Cannot Establish
The review’s findings depend on its selection of 18 claims from a 50-indicator list and on the author’s chosen verification categories. The source material does not provide a full independent reproduction of every result or a shared method for deciding when a claim qualifies as verified. Some underlying claims remain company-reported, single-sourced or disputed, and the article does not resolve all of those disagreements.
It is also unclear how consistently the tracker applies its indicators, particularly where a present-day research result is grouped under a much larger future milestone. Nor does a mathematical achievement answer whether a model is autonomous, reliable across fields or capable of replacing human expertise. The source reports no household robots at scale or visible US unemployment-rate break, but does not specify measures that would define those thresholds. The evidence reviewed does not settle when, or whether, AGI or ASI has arrived.
The Next Tests to Watch
The next useful evidence will be results that can be checked beyond a company announcement or a checklist entry: published methods, independently reviewed work, reproducible evaluations and clear accounts of how much human direction a system required. For claims involving research acceleration, comparisons should state the task, time period and baseline, rather than relying on broad descriptions of faster work.
Thompson’s tracker is described as a continuing public list, and the source article reports its status as of October 26, 2026. Its future updates may change which indicators are marked achieved or partial. Until definitions and verification standards are clearer, readers should treat the tracker as one researcher’s framework for monitoring developments, not as a consensus declaration that superintelligence has arrived.
Key Questions
What did the October 26 review find?
ThorstenMeyerAI.com said its check of 18 selected claims classified seven as verified, five as vendor claims, three as single-source claims and three as disputed or contradicted. The categories reflect the review’s own assessment.
How many signs does Thompson say have been achieved?
The tracker was reported as listing two achieved, 33 partially achieved and 15 not achieved as of October 26, 2026. Thompson’s partial category includes enabling work and funded programmes, so it does not mean those signs are nearly complete.
Does the tracker prove that AI is already superintelligent?
No. The tracker presents Thompson’s assessment, and the review says the meaning of “ASI” varies across frameworks. The evidence it checked includes genuine advances, but it does not establish that current systems meet a broadly accepted superintelligence standard.
What kinds of AI advances did the review find strongest?
The review said the best-supported examples clustered in areas where evidence can be checked, including mathematical proofs, a physics calculation and regulatory filings. It also noted caveats about human involvement and the scope of individual results.
What remains uncertain about AI’s effects?
The review does not establish when AGI or ASI might arrive, or how quickly AI will affect jobs and physical-world work. It reports no visible break in the US unemployment rate and no household robots at scale, but those observations do not settle future impacts.
Source: ThorstenMeyerAI.com
