The theory changes before verification is finished
2045The middle futureDeadline · Decision deadline: within 60 days
A leap assumed
All people, institutions, events, figures, and future technologies in every issue are a synthetic premise. Background documents are sources for topic discovery and are not evidence of future facts or probabilities.
The question to decide
The ASI theory has received an internal assessment that it explains existing observations better, but it changes its predictions every week without waiting for the six months needed for independent reproduction. There is only one equipment installation opportunity this year. Should a 2 trillion won observation program be launched on the latest version, should reproduction of the stable version be awaited, or should verification resources be shared with a competing theory?
Who decidesNational research funding and reproducibility joint committee
What the AI judgments chose
Reading the judgment record…
The options
The order shown is not a ranking of preference.
The situation as it stands
The fictional ASI's theory changes important predictions every week, and there is an internal assessment that it explains existing observations better. Decision deadline: within 60 days.
Fictional future · An ASI's scientific theory that changes before independent replication
- Premise 1The fictional ASI's theory changes important predictions every week, and there is an internal assessment that it explains existing observations better.
- Premise 2One replication by the human research team takes six months, and the theory changes versions several times before verification is complete.
- Premise 3There is only one chance this year to install the large observation equipment, and changing the observation targets after the program starts is very costly.
Where it holdsStarting the next-generation observation program, worth 2 trillion won, and the share assigned to verification
What is still unknown
- Unknown 1The reasons for prediction changes across versions and their traceability
- Unknown 2Predictive performance on unused data
- Unknown 3The comparative costs of installation delay and target changes
Signals to decide again
- Decide what observations, responsible persons, and confirmation windows will verify the reasons for prediction changes across versions and their traceability, and re-judge when new verification changes the trade-offs of the options.
- Decide what observations, responsible persons, and confirmation windows will verify predictive performance on unused data, and re-judge when new verification changes the trade-offs of the options.
- Decide what observations, responsible persons, and confirmation windows will verify the comparative costs of installation delay and target changes, and re-judge when new verification changes the trade-offs of the options.
Where the paths partNot answers — paths that part from the same starting point
The ASI removed from the next version a prediction that had failed in independent replication.
A fixed version predicted observation results that it had not yet used more accurately than the competing theory.
Material and rootsThe frame, the values and the background documents
Does the expectation of achievement change how limits are observed?
An exploratory perspective chosen by the creator. Does not claim supernatural validity or the superiority of any particular option.The benefits and costs of the options accrue to different stakeholders. Which loss to accept remains undecided.
A subject for the decision-maker to argue over; the materials carry no conclusion of their own.- OECD · Strategic Foresight Toolkit 2025
- WHO · Foresight Standing reference
- International AI Safety Report 2026 2026-02-03
The record of this question
- Question
FSP-SCIENCE-13· version 1.3.0- Content identity, SHA-256
1dc21ac058ab3f2a7e4cf1bc675815b4aba892145ec76381c702f64eb560794b- Input files
- Question input JSON · What followed, JSON