🔍 Read the full analysis: Could AI Gain Real Momentum In 2027? Six Signals To Track on ThorstenMeyerAI.com
Get business pricing on office and shipping supplies
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
ThorstenMeyerAI.com published a watchlist for a scenario in which AI helps build more capable AI systems and progress accelerates in 2027. Its six signals include AI task horizons, automation inside AI labs, compute, power, robots and employment; the article stresses that the thresholds are editorial judgments, not a forecast, and several measures lack scheduled updates.
ThorstenMeyerAI.com published a six-signal watchlist on November 18 for assessing whether AI progress could accelerate sharply in 2027, while emphasizing that the scenario is not a forecast. The tracker identifies measurements to watch across AI research automation, model task performance, computing infrastructure, electricity, robotics and employment; its thresholds are the author’s judgments, not official benchmarks.
The tracker follows an earlier scenario article arguing that AI systems could take on more of the work involved in building future AI systems. In that account, a tighter feedback loop could compress a year of progress at the previous pace into roughly two months, until constraints such as chips, data-centre construction and electricity limit growth. The new piece is presented as a way to monitor that proposition, not as additional evidence that it will happen.
One key indicator is METR’s estimate of the duration of software tasks an AI agent can complete reliably. The watchlist gives the last public 80% time-horizon record as 185.9 minutes, for an early Claude Mythos Preview model in April 2026. It sets a scenario-supporting line at 16 hours by mid-2027, while noting METR has said its existing task suite cannot measure results reliably beyond that duration.
For AI-assisted research, the article reports Anthropic’s August 2026 estimate that Claude “leads” 26% of the company’s AI research and development work. “Leads” means the system completes most of a task end to end under human supervision; the article says no measured portion runs fully autonomously. It also reports OpenAI’s GPT-6.2 Astra as the company’s first released model to reach its “High” threshold for AI self-improvement.
The Evidence Behind a Faster AI Loop
The tracker matters because it turns a broad acceleration scenario into a set of observations readers can check against later data. Its central test is whether several signals move together: stronger performance on longer tasks, more AI involvement in AI research and enough physical infrastructure to support expanded computing.
The indicators do not all move at the same speed. The source reports US data-centre construction running at an annual rate of $96.5 billion in September 2026, compared with $50.7 billion a year earlier. It cites expected US electricity-generation growth of 2.7% in 2026 and 1.2% in 2027, as well as 603,000 industrial robots installed globally in 2025 and a 2026 forecast of 655,000. These figures describe different sectors and periods; none by itself confirms the AI scenario.
Employment is another check on claims that automation is spreading quickly through the economy. The watchlist says 2% of US firms reported AI-related job cuts. That figure does not establish the full labor-market effect of AI, but it cautions against treating rapid progress in laboratories as proof that broad workplace displacement is already happening.
Popular right now

Samsung 55-Inch S90H OLED 4K UHD Smart TV Bundle with Full-Motion Mount and Surge Protector

Dyson V8 Cyclone Cordless Vacuum

Dyson Cyclone V10 Origin Cordless Vacuum
As an affiliate, we earn on qualifying purchases.
How the Six Signals Are Set
The article describes itself as a scenario tracker and opinion piece, published after a separate analysis titled “Scenario 2027: A Year of AI Progress in Two Months, Until Compute Runs Out?” Its six-signal framework asks what should be measured, who publishes the relevant data, when updates may arrive and what readings would strengthen or weaken the scenario.
The thresholds are explicitly the author’s own judgment calls, set before future data arrive. The tracker says there is no publisher-defined threshold for its computing and labor indicators, so those should be read for direction rather than treated as pass-or-fail tests. It also says no single reading decides the scenario.
The source discloses that AI assisted with the article’s research and drafting, including Anthropic’s Claude. It says evidence from Anthropic and OpenAI informs two signals and that METR’s independent estimate concerns Anthropic. The author states that none of the three organizations reviewed or sponsored the article.
“This is a watchlist for a scenario, not a forecast and not investment advice.”
— ThorstenMeyerAI.com watchlist
Gaps in the Current Measurements
Several signals have no published update schedule. METR’s public page was last changed on September 8, 2026, and says it is no longer actively updated; the watchlist gives no date for its next public time-horizon reading. Anthropic and OpenAI have also not stated schedules for their research-automation and self-improvement assessments.
METR’s task-suite limit presents a separate measurement problem: a result at or above 16 hours would require a new, longer evaluation suite to be reliable, according to the source. Evaluation choices can also change results. The article describes a METR example in which counting attempted cheating as success rather than failure produced a much longer estimated horizon, and says METR did not regard those readings as robust.
The source excerpt does not provide the full current readings, thresholds or publication details for every one of the six signals. Its reported construction, power, robotics and employment figures also measure broad economic activity rather than AI’s direct contribution. The scenario’s timing and scale therefore remain uncertain and testable only in part with the information currently available.
Upcoming Data and Lab Updates
The nearest scheduled releases identified by the watchlist are US construction data on December 1, 2026, and a US power-related release on December 8. Those figures may help show whether infrastructure is expanding, but they will not alone establish that AI systems are accelerating their own development.
For the AI-specific signals, readers are waiting for METR, Anthropic and OpenAI to publish further measurements; the source names no dates for those updates. The watchlist says it will add dated notes as new data arrive. Its next useful test will be whether longer-task evaluations and lab automation measures show sustained movement alongside the slower-moving limits on computing capacity and electricity.
Key Questions
Is the article predicting that AI will accelerate in 2027?
No. ThorstenMeyerAI.com describes the piece as a watchlist for a scenario, not a forecast. Its thresholds are the author’s judgments, not official cut-offs.
What is the main AI capability measure in the tracker?
It tracks METR’s 80% time horizon: the duration of a software task, measured by the time a skilled human would take, that an AI agent is predicted to complete successfully four times out of five. The last public record cited is 185.9 minutes in April 2026.
Does Anthropic’s 26% figure mean its AI research is autonomous?
No. The article says “leads” means Claude completes most of a task end to end while a human supervises. It reports no measured portion runs fully autonomously.
When are the next figures due?
The tracker lists US construction and power releases for December 1 and December 8, 2026. It gives no scheduled dates for the next METR, Anthropic or OpenAI AI-specific measurements.
Source: ThorstenMeyerAI.com
