50 Signs Of Superintelligence: Where AI Fits On The Road To ASI
AIThis post was created with the assistance of artificial intelligence (AI).

🔍 Read the full analysis: 50 Signs Of Superintelligence: Where AI Fits On The Road To ASI on ThorstenMeyerAI.com

Buying for a business?Offer from Amazon

Get business pricing on office and shipping supplies

  • Business-only prices and quantity discounts
  • Tax-exempt purchasing
  • Multiple users, one account, clear invoices
As an affiliate, we earn on qualifying purchases.

TL;DR

A public checklist by AI consultant Alan D. Thompson lists 50 signs of artificial superintelligence, but its status labels include early research and funded programs, not just completed achievements. A review of 18 prominent claims found a mix of independently checked results, company-reported claims and disputed or unsupported assertions; the definition of “ASI” also differs from major labs’ definitions.

A review of 18 prominent claims in AI consultant Alan D. Thompson’s 2026 montage of signs of artificial superintelligence found that some are backed by research or official records, while others remain company claims, rely on one source, or are disputed. Thompson’s tracker marks 2 of 50 indicators achieved, but its “partially achieved” category can include funded programs and preparatory work, making the list a watchlist rather than a direct measure of how close AI is to superintelligence.

Thompson, an Australian AI consultant who publishes The Memo newsletter, maintains a public 50-indicator ASI checklist covering developments from AI-assisted mathematics to much more speculative milestones. In the tracker’s October 26 status grid, 33 indicators are partially achieved and 15 are not achieved. Thompson’s August correction note says partial status includes “documented material progress” such as funded programs, research infrastructure, formal scientific missions and enabling work. It does not mean a milestone is nearly complete.

The source article’s review checked 18 of the montage’s stronger claims against papers, researcher accounts, company posts and other primary material. It classified 7 as verified, 5 as vendor claims, 3 as relying on a single source and 3 as disputed or contradicted. The verified examples cluster in areas where an outcome can be examined, including mathematical work and formal records. These labels describe the review’s findings; they are not an independent audit of every item on Thompson’s 50-point list.

Examples include AI contributions discussed by mathematicians Terence Tao, Don Knuth and Tim Gowers, each with qualifications around coaching, prior literature or the type of problem. Other claims are less independently established. For example, the source describes a claim about a set of results attributed to GPT-6 Astra as a vendor claim: Lean certificates are public, but the results were not independently reproduced in the material reviewed. The distinction matters: a public artifact can support scrutiny without proving every broader claim made about a model.

At a glance
reportWhen: Published October 26, 2026; tracker sta…
The developmentA review of Alan D. Thompson’s 50-indicator ASI tracker and 2026 video finds uneven evidence behind headline claims and a broad definition of superintelligence.
Thorsten Meyer AI
Definitions

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.

2indicators achieved (maths proof, conjecture)
33partially achieved, incl. funded programmes
15not achieved
99%Thompson's own AGI countdown
  1. Median human50th percentileThompson: AGIa machine at the level of an average (median) humanDeepMind: Competent AGIat least the 50th percentile of skilled adults
  2. Expert90th percentileThompson: ASIexpert-human level in practically any fieldDeepMind: Expert AGIat least the 90th percentile of skilled adults
  3. Virtuoso99th percentile—DeepMind: Virtuoso AGIat least the 99th percentile of skilled adults
  4. Beyond every human100%—DeepMind: Superhuman (Level 5)outperforms 100% of humans
  5. 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 story

The disagreement is not only about whether a model can solve a hard problem. It is also about what counts as superintelligence. Thompson defines ASI as a machine performing at expert-human level in practically any field. Under that definition, strong performance in some expert tasks may be treated as evidence toward ASI, even when capability is uneven or human guidance was involved.

Other frameworks set a higher bar. Google DeepMind’s 2023 “Levels of AGI” scale distinguishes competent, expert, virtuoso and superhuman performance. Its June 2026 report describes superintelligence in relation to systems 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 standards put different weight on breadth, autonomy and performance beyond people.

For readers, the practical consequence is that claims such as “early ASI” are difficult to compare without a stated yardstick. A model’s contribution to a proof may be meaningful evidence of progress in mathematics, but it does not by itself establish broad, autonomous ability across fields. Specific task results and the conditions under which they were produced are more informative than a label alone.

De'Longhi La Specialista Arte Evo Semi-Automatic Espresso Machine with Grinder and Cold Brew

De'Longhi La Specialista Arte Evo Semi-Automatic Espresso Machine with Grinder and Cold Brew

Breville Barista Express Espresso Machine with Grinder, Black Truffle

Breville Barista Express Espresso Machine with Grinder, Black Truffle

Ninja Air Fryer XL, 6.5 Qt, 6-in-1, MaxCrisp, Dark Grey (AF181)

Ninja Air Fryer XL, 6.5 Qt, 6-in-1, MaxCrisp, Dark Grey (AF181)

As an affiliate, we earn on qualifying purchases.

How the tracker defines progress

Thompson’s October video, “ASI: A better world – 2026,” is a fast montage of checklist entries from January to October. The tracker names Claude Mythos Preview and GPT-6 Astra as “early ASI” systems, a characterization that should be understood as Thompson’s assessment, not a consensus finding or a formal certification by the companies involved.

The list ranges from claims about recursive code self-optimization to milestones far removed from present systems. The source article notes that two exoplanet candidates found with Claude Code were placed under an indicator about the first planet other than Earth being colonized. That illustrates how partial status can register activity related to an indicator without meaning the stated milestone has occurred.

Thompson also publishes a separate, conservative AGI countdown that he puts at 99%. He says the final step depends on a robot completing multi-step fine assembly, such as putting together furniture, at average-human reliability. That is a separate estimate based on his own criteria, not a measured probability that AGI will arrive soon.

““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 for his tracker

What the evidence cannot settle

The source article’s sample of 18 claims is not a full verification of all 50 indicators, and its verdicts depend on the sources available to the reviewer. “Verified” does not mean a claim establishes ASI; it means the stated result was supported in the review by an independent expert or official document. Vendor-reported performance, meanwhile, may be genuine but has not necessarily been reproduced independently.

It remains unclear how many of the tracker’s partially achieved indicators correspond to completed, independently validated capabilities rather than preparatory activity. Nor does the material establish that the models Thompson calls “early ASI” meet a common definition of AGI or superintelligence. The source also reports no household robots at scale, no verifiably billion-dollar AI-run company and no visible break in the U.S. unemployment rate, but those observations alone cannot establish what AI’s broader economic effects will be.

Claims about recursive self-improvement also need careful scope. The source reports Anthropic’s estimate that AI has made its research and development faster, but not by a factor of two, and that its best internal model scored 62.8% on a test where the company said full replacement of its researchers would require at least 85%. Those figures are company-reported measures, not evidence that AI can independently replace an entire research team.

What to watch in later checks

The next useful test is whether claimed capabilities can be reproduced independently, with clear reporting of the model, tools, human guidance and evaluation method. For mathematical results, that means checking the proof and its assumptions; for broader claims, it means testing performance across tasks rather than relying on a single demonstration or company-selected metric.

Readers can also track changes to Thompson’s public checklist, including whether partial entries move to achieved and what evidence supports the change. The tracker’s status should be read alongside the wording of each indicator and the definitions used by other researchers. The source material does not identify a scheduled independent reassessment or a common industry benchmark that will settle the competing claims.

Key Questions

What does the 50-signs ASI tracker report?

As of October 26, 2026, it lists 2 indicators as achieved, 33 as partially achieved and 15 as not achieved. Thompson says partial status can include funded programs and other enabling work.

Does “partially achieved” mean ASI is close?

No. Thompson’s definition includes research infrastructure, funded programs and preparatory work. It does not necessarily mean the milestone itself is near completion.

Are today’s leading AI models confirmed to be early ASI?

No shared confirmation is established in the source material. Thompson calls two models “early ASI,” but that is his assessment and rests on a broader definition than some lab frameworks use.

How strong was the evidence in the claims reviewed?

The source article classified 7 of 18 claims as verified, 5 as vendor claims, 3 as based on a single source and 3 as disputed or contradicted. The review did not cover every indicator on the full checklist.

Source: ThorstenMeyerAI.com

This content is for general information only and is not financial, tax or legal advice. Consult a qualified professional for decisions about your money.
You May Also Like

The Hyperscaler Prisoner’s Dilemma: Why I Keep Buying Nvidia

Analysis of why major hyperscalers and investors continue purchasing Nvidia stock despite industry challenges and market volatility.

2026’S Most Popular Mesh WiFi Systems For Home Use

Discover the most popular mesh WiFi systems of 2026, including WiFi 6 and WiFi 7 options, their features, and which suits your home best.

Disk Is the Contract: Inside Threlmark’s Local-First Architecture

Threlmark introduces a local-first, file-based architecture that makes project data portable, inspectable, and restartable without relying on a server or database.

From Sensors To Software Autonomy: AI’s Transformative Power

New developments show European agencies contracting for independent ISR software, marking a shift toward sovereignty in sensor-to-software AI systems.