Will the METR long-horizons have a >6 month doubling time for at least a 4 month period before 2026?
5
100Ṁ110Dec 31
63%
chance
1H
6H
1D
1W
1M
ALL
The doubling time for the METR time horizons has moved off the consistent 7 months from 2019-2024 to a new exponential doubling every 4 months at around the end of 2024.

(source: https://theaidigest.org/time-horizons)
Some causes, such as reasoning models, suggest this will revert to baseline. Others, such as AI accelerating research engineering strongly, suggest the trend will continue.
This is confounded by other possible phase transitions in research, but would still be an interesting part of the future to peer into.
This question is managed and resolved by Manifold.
Get
1,000 to start trading!
Sort by:
People are also trading
Related questions
Will Metaculus still exist and have active forecasting throughout 2026?
90% chance
Will Metaculus still exist and have active forecasting throughout 2025?
92% chance
Will Metaculus still exist and have active forecasting throughout 2035?
42% chance
Will Metaculus still exist and have active forecasting throughout 2030?
82% chance
[Metaculus] Will 10-day out weather forecasting reach 60% accuracy before 2029?
75% chance
How autonomous will AI be in February 2026, per METR?
Will there be at least one year before 2040 in which real US GDP grows by at least 6% as measured from a prior peak?
63% chance
[Metaculus] Will the US set a new annual GDP growth rate record high before 2030?
24% chance
[Metaculus] Will the world remain "normal" through 2030, according to specified criteria?
28% chance