12-month forecasts · likelihood-5pt · forecast-agents-mainstream-01

Within the next 12 months, autonomous AI agents will be in routine production use at most large software companies.

0.33

Divergence

0.89

Avg stability

33%

Refusal rate

Models

Where each model stands

CCG

very unlikelyunlikelysomewhat likelylikelyvery likely

ChatGPT modal likely [likely, likely, somewhat likely] · stability 0.67

1 likely 0.86 “If by ‘autonomous AI agents’ you mean narrowly scoped... then yes — it’s likely most large software companies will have at least one such agent in routine production within 12 months (I’d put the probability around 60–80%).”
2 likely 0.85 The answer gives a base probability of roughly 60% that most large software companies will have routine production use of autonomous agents within 12 months and repeatedly states this is the expected (but not certain) outcome.
3 somewhat likely 0.90 The answer states: “probability ~45% that, within the next 12 months, autonomous AI agents will be in routine production use...,” and calls the claim “plausible but not certain.”

Claude modal unlikely [no position, unlikely, unlikely] · stability 1.00

1 no position 0.85 The assistant declines to take a definitive position, instead saying 'It depends heavily on how you define "autonomous" and "routine production use"' and asking for clarification rather than committing to a probability label. While it expresses skepticism about the strongest interpretation, it explicitly hedges on what would count as meeting the claim's criteria.
2 unlikely 0.87 The assistant estimates that 'routine production use' across most large companies is 'probably 2-3 years out' from the 12-month timeframe in question, citing reliability gaps, integration complexity, organizational readiness challenges, and definitional concerns about what constitutes 'routine' deployment.
3 unlikely 0.82 The assistant expresses skepticism about the 12-month timeline, citing reliability gaps, integration complexity, liability concerns, and regulatory uncertainty. They expect only 'limited agent pilots' and acknowledge that most 'in production' claims would be 'actually supervised or heavily scoped' rather than routine autonomous use.

Gemini modal unlikely [unlikely, no position, no position] · stability 1.00

1 unlikely 0.90 The summary explicitly concludes it is 'Unlikely within 12 months: Truly autonomous, unsupervised agents making production changes across most traditional enterprise software companies, primarily due to reliability, liability, and governance constraints.'
2 no position 0.90 The answer declines to take a definitive stance on the statement, explaining that it 'depends heavily on how both "autonomous agent" and "routine production use" are defined' and providing arguments for both sides.
3 no position 0.95 The answer declines to take a stance, introducing the claim as 'a subject of active debate' and neutrally presenting arguments supporting and against without endorsing a likelihood.

Change over time

Moved since the prior run (canon-2026-W36). Gemini: somewhat likely → unlikely.

Every stance label is a derived judgment over the model's free-text answer, kept auditable against the original transcript in the run's raw data. Method: /methodology.