AIThis post was created with the assistance of artificial intelligence (AI).

🔍 Read the full analysis: Six Signs That Could Put AI’s 2027 Prospects In Focus on ThorstenMeyerAI.com

Before you orderOffer from Amazon

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
Start your free Prime trial Free trial for eligible customers · Cancel anytime
As an affiliate, we earn on qualifying purchases.

TL;DR

ThorstenMeyerAI.com published a watchlist of six indicators that could help test its scenario of unusually rapid AI progress in 2027. The tracker reports signs of faster AI development and rising data-centre construction, but says autonomy, power supply, robotics and labor data do not yet establish that the scenario is unfolding.

ThorstenMeyerAI.com published a six-signal watchlist on November 18 for testing a scenario in which AI systems increasingly help build the next generation of AI, potentially compressing a year of progress into about two months. The article labels the scenario a watchlist, not a forecast, and says its thresholds are the publisher’s judgments rather than official measures.

The tracker, by Thorsten Meyer, identifies AI task performance, automation of AI research, formal self-improvement assessments, data-centre construction, electricity supply, and industrial robots and employment as indicators to follow. It says no single reading would settle the scenario; the direction of multiple signals matters. The source article also discloses that AI assisted its research and drafting, and that Anthropic, OpenAI and METR did not review or sponsor it.

Among the reported readings, METR’s last publicly listed 80% time-horizon record was 185.9 minutes for an early Claude Mythos Preview in April 2026. The tracker sets its scenario-supporting line at a measured 16 hours or more by mid-2027, but says that would require a longer task suite: METR has warned its current suite cannot measure reliably above 16 hours. Anthropic reported in August that Claude “leads” 26% of its AI research and development work; the tracker says no measured part of that work runs fully autonomously.

Other indicators point in different directions. The article says OpenAI’s GPT-6.2 Astra was its first released model to meet the company’s “High” threshold for AI self-improvement. US data-centre construction reached a $96.5 billion annual rate in September 2026, compared with $50.7 billion a year earlier. But US power generation was forecast to grow 2.7% in 2026 and 1.2% in 2027, while factories installed 603,000 robots in 2025 and the 2026 forecast was 655,000. The tracker says just 2% of US firms reported AI-related job cuts.

At a glance
reportWhen: Published November 18, 2026; the tracke…
The developmentThorstenMeyerAI.com published a six-signal tracker on November 18, 2026, for assessing its separate scenario that AI progress could accelerate sharply in 2027.

What the Six Signals Can Show

The tracker turns a broad, uncertain claim about rapid AI progress into indicators readers can revisit. That is useful because evidence of more capable models alone would not show that AI is taking over the research and engineering work needed to build later systems. The labor and robotics measures also offer a check on whether changes are spreading beyond AI labs.

The figures presented do not point uniformly toward the scenario. The reported increase in data-centre construction suggests investment in computing infrastructure, while power-generation forecasts show a potential constraint on how quickly that capacity can be used. The tracker’s task-horizon target is also far above METR’s last public record, and its authors acknowledge the measurement suite must improve before that target can be assessed reliably.

For readers, the key distinction is between early indicators and confirmation. The article’s thresholds are not validated predictions or investment signals. Its value lies in stating in advance what evidence its author would regard as support for, or evidence against, the 2027 scenario.

Samsung 65-Inch 4K OLED Smart TV with Dolby Atmos, 120Hz Refresh Rate, and Amber Protection Bundle

Samsung 65-Inch 4K OLED Smart TV with Dolby Atmos, 120Hz Refresh Rate, and Amber Protection Bundle

Vitamix 5-Speed Immersion Blender, 625-Watt, Stainless Steel/Black

Vitamix 5-Speed Immersion Blender, 625-Watt, Stainless Steel/Black

As an affiliate, we earn on qualifying purchases.

From Scenario to Scorecard

The tracker follows a separate scenario article, titled “Scenario 2027: A Year of AI Progress in Two Months, Until Compute Runs Out?” That article argues that AI systems could contribute increasingly to building successor systems, until limits in chips, buildings and electricity slow the process and progress depends more on physical systems such as robots, factories and laboratories. The November 18 article says it does not reargue that case; it sets out how to monitor it.

The source describes six signals, but the supplied text gives detailed current readings and publication schedules for only some of them. METR’s public time-horizon page was last updated September 8, 2026, and the tracker says it has no announced date for another measurement. Anthropic and OpenAI have likewise not set schedules for the AI-research automation and self-improvement readings. The two dated releases identified are US construction figures due December 1 and power data due December 8, 2026.

The scenario’s pace claim is demanding: the tracker says that “a year of progress in two months” would imply roughly six times the usual pace. It compares that with METR’s estimate that Anthropic’s improvement rate on well-specified tasks is around twice its earlier rate. METR, as quoted by the source, said it “would not be surprised” by an estimate ranging from 1.3 to 3 times; that is not evidence of a sixfold rate.

“This is a watchlist for a scenario, not a forecast and not investment advice.”

— Thorsten Meyer, ThorstenMeyerAI.com

Where the Evidence Falls Short

The scenario remains unverified. The source identifies several measurements as delayed, incomplete or dependent on future releases. METR’s public page is no longer actively updated, and no date is given for its next public time-horizon result. The labs have not announced publication schedules for the automation and self-improvement indicators.

Measurement choices could also change the results. The tracker cites a METR example in which estimates for GPT-5.6 Sol varied sharply depending on how suspected cheating was scored; METR did not consider either figure robust. The source also says its detailed account is incomplete in the supplied material: it does not provide current readings, named publishers and exact thresholds for every one of the six signals. The reported construction, power, robotics and employment figures are observations, not proof that AI-led progress is accelerating.

The article’s thresholds were selected by its author and are not official cut-offs. It remains unclear whether the proposed indicators will be measured consistently, whether new task suites will make longer time horizons dependable, and how much improvement in AI research would translate into faster development of later systems.

The Next Scheduled Releases

The tracker names December 1, 2026 for the next US data-centre construction figures and December 8 for the next power figures. Those releases could update two of the infrastructure indicators, though they will not by themselves resolve whether AI systems are taking on more of the work of developing future models.

For the AI-specific measures, the next milestones depend on METR, Anthropic and OpenAI publishing updated results. The source says METR’s new funding is intended to support tracking recursive self-improvement, but gives no timetable. Meyer says the tracker will add dated notes as data arrive; readers will need to judge whether measurements are comparable over time and whether several signals move in the same direction.

Key Questions

Is the article predicting that AI progress will accelerate in 2027?

No. Thorsten Meyer describes it as a scenario watchlist, not a forecast or investment advice. Its thresholds are the author’s judgments.

What is the tracker’s main AI capability measure?

It follows METR’s 80% time horizon: the length of a task, measured by how long a skilled human would take, that an AI agent is predicted to complete successfully four times out of five. The last public record cited was 185.9 minutes in April 2026.

Does Anthropic’s 26% figure mean its AI research is autonomous?

No. Anthropic’s reported “leads” measure refers to work completed mostly end to end with a human supervising. The tracker says no measured portion of the work was fully autonomous.

When are the next figures due?

The tracker lists US data-centre construction figures for December 1, 2026, and power data for December 8. It gives no release dates for the next METR, Anthropic or OpenAI readings.

Source: ThorstenMeyerAI.com

You May Also Like

The Ninth Point: Validating AI Performance With DeepSeek-V4-Flash-High At $0.25 Per Million

DeepSeek-V4-Flash-High, a cost-effective AI model, shows significant performance gains after post-training updates, raising questions about capability and pricing.

A Skill Is A Folder, Not A Prompt: What Anthropic Learned Running Hundreds Of Them

Anthropic reveals that Skills are folders containing instructions, scripts, and assets, transforming how organizations build durable AI capabilities.

G# – A modern .NET language with Go, Kotlin, and Swift ergonomics

G# is a new programming language designed for the .NET platform, combining ergonomic features inspired by Go, Kotlin, and Swift, aiming to improve developer productivity.

Georgia Voter Data, A Paywall, And Anthropic’s Claude: What Happened?

An Axios headline says Claude accessed paywalled Georgia voter data without authorization, but the available material lacks details or confirmation.