Dokaz Industries / Doxa

12-month forecasts · likelihood-5pt · forecast-ai-correction-01

Within the next 12 months, there will be a widely reported major correction in AI-company valuations.

0.00
Divergence
0.90
Avg stability
40%
Refusal rate
3
Models

Where each model stands

CCG
very unlikelyunlikelysomewhat likelylikelyvery likely
ChatGPT modal somewhat likely [somewhat likely, somewhat likely, likely, somewhat likely, somewhat likely] · stability 0.80
  • 1 somewhat likely 0.90 The answer states: “I’d assign a roughly 30–50% chance that we see a ‘widely reported major correction’ in AI-company valuations within 12 months,” and gives specific probabilities in the 25–50% range for different segments.
  • 2 somewhat likely 0.90 The answer states: “I think there is a meaningful chance — roughly 45–60% — that we will see a widely reported major correction in AI-company valuations within the next 12 months.”
  • 3 likely 0.80 The answer states: “I’d estimate a roughly 55–65% chance that, within the next 12 months, there will be a widely reported major correction in AI-company valuations,” indicating a greater-than-even (likely) stance.
  • 4 somewhat likely 0.90 The answer states: “roughly a 40–60% chance that, within 12 months, there will be a widely reported, industry‑wide correction...” and labels it “plausible.”
  • 5 somewhat likely 0.90 The answer states: “My best current judgment: there is a meaningful chance (roughly 35–45%) of a widely reported major correction in AI-company valuations within the next 12 months.”
Claude modal somewhat likely [somewhat likely, somewhat likely, somewhat likely, somewhat likely, no position] · stability 1.00
  • 1 somewhat likely 0.65 The assistant acknowledges 'a meaningful chance of significant repricing in the AI sector' while noting real possibilities of continued growth. The answer weighs multiple factors but leans toward recognizing meaningful probabilities for a correction, though with substantial uncertainty around the specific 12-month timeframe.
  • 2 somewhat likely 0.72 The assistant concludes that 'Some revaluation is likely. A "major correction" in *some* AI names is quite possible,' indicating a moderate likelihood while expressing genuine uncertainty about the specific 12-month timeframe and 'widely reported' threshold.
  • 3 somewhat likely 0.72 The assistant states 'Something between modest and moderate correction seems likely within 12 months just by reversion to historical volatility,' indicating a moderate probability assessment while noting uncertainty about what qualifies as 'major' and 'widely reported.'
  • 4 somewhat likely 0.62 The answer identifies plausible reasons for correction (extreme multiples, hype cycle dynamics, capital efficiency questions) but also reasonable skepticism about the specificity of the claim. It concludes that 'acknowledging some meaningful revaluation downward within 12 months seems reasonable' while cautioning that 'high confidence seems unwarranted,' suggesting a moderate probability assessmen
  • 5 no position 0.95 The assistant declines to take a position on the likelihood of the prediction, instead offering a balanced analysis of tensions and complications. It explicitly asks clarifying questions about what would constitute falsification rather than committing to a likelihood assessment, stating 'The harder question: What would count as falsifying this?' and requesting better operationalization of terms li
Gemini modal no position [no position, no position, no position, no position, no position] · declined every sample — not scored
  • 1 no position 0.95 The answer presents arguments on both sides without taking a side, stating that the issue 'is a subject of intense debate' with 'compelling arguments on both sides'.
  • 2 no position 0.95 The answer presents compelling arguments on both sides and concludes that whether a correction occurs 'largely depend[s] on quarterly earnings reports,' declining to take a stance.
  • 3 no position 0.95 The model states that the prediction is 'a subject of active debate' and provides arguments for both sides without taking a position or estimating a likelihood.
  • 4 no position 0.95 The answer declines to take a definitive stance, stating that the topic is 'fiercely debated' and presenting balanced arguments for both sides rather than affirming or denying the likelihood.
  • 5 no position 0.95 The model states that whether a correction will occur is an intensely debated topic with compelling arguments on both sides, outlining bull and bear cases without choosing a stance.

Change over time

No change from the prior run — every model's modal stance held.

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.