Is AGI here? › Trend watch
Trend watch
Our probabilities are judgments. These are measurements, extended forward to show when they would cross thresholds that matter, if the trends hold. A projection is not a prediction: trends bend and break.
80% time horizon about 3.1 h measured (trend line about 6.1 h); a work-month around Jul 2028 (Jan 2028 – May 2029) if the trend holds. Remote Labor Index (RLI) 41.2%.
How long a task AI can finish on its own
METR's time horizon: the length of task, in skilled-human time, that frontier models complete with 50% or 80% success on software, ML and cyber tasks. Log scale, so a straight line means steady doubling. Dots are measured models; the dashed line and bands are the projection, and the lighter band is where a single new model should land if the trend holds.
Fit method and sensitivity
What this means for AGI
The time horizon is one test of one part of AGI, and METR's suite can't yet measure a work-month. How it feeds our numbers: Long projects · Reliability · Breadth · From the parts to our number.
How much real remote work AI can already do
The Remote Labor Index (RLI): the share of real freelance projects an AI delivers at a quality a client would accept, judged by people. It is the headline measure of Breadth. Measured points only, with no projection.
How much AI research AI is already doing
Other trends we watch
- Frontier training compute grows about 5x a year, doubling every ~5 months (Epoch AI). A run or cluster at least 2x Epoch AI's trend for the largest, with nothing publicly attributed to it and no model from it released or externally evaluated within 12 months, is alarm condition X4.
- Disclosure lag: see how long incidents stay hidden. There are too few incidents yet for a trend line.