Kvíz spojený s průvodcem · Střední úroveň
The Track Record of AI Predictions Quiz
Tests whether readers can assess AI forecasts by checking dates, defined tasks, assumptions and outcomes, and separate demonstrations from adoption.
Otázka 1 z 8
Which forecast is easiest to assess later?
Náhled každé otázky8 otázky
Odpovědi a vysvětlení se zobrazují, jak budete v kvízu.
- Which forecast is easiest to assess later?
- AI will change everything soon
- A named system will reach a stated benchmark by a dated deadline under specified test conditions
- Machines will be smarter than people someday
- Technology will keep improving
- Simon’s 1965 forecast said machines would be technologically capable of doing any work a person could do within twenty years. Which outcome would most directly test that claim?
- Whether all human workers had been replaced by machines by 1985
- Whether computer use increased in some offices
- Whether broad technical capability across human work had been demonstrated by 1985
- Whether AI research budgets grew by the deadline
- Why should an evaluator preserve a forecast’s original wording?
- To keep later reinterpretation from changing what was claimed
- To make the prediction come true
- To avoid stating the date
- To remove uncertainty from the outcome
- A system beats people on one benchmark. What does that alone establish?
- It can perform every human job
- It is generally intelligent
- It will be adopted widely
- It performed well on that benchmark under its test conditions
- Which information belongs in a prediction ledger?
- Only predictions that succeeded
- Dated claims, assumptions, outcomes and unresolved cases
- Only company press releases
- Forecasts without deadlines
- Why is “human-level AI soon” hard to score?
- It names a precise task and deadline
- It can be verified from one model answer
- It leaves “human-level” and “soon” undefined
- It is a forecast about hardware only
- How should a capability demonstration be distinguished from adoption?
- A demo proves most organizations use it
- Adoption is irrelevant to forecasts
- All demos are products
- A demonstration shows a result in a test; adoption concerns actual use at scale
- A forecast misses its deadline. What can that show about every future AI forecast?
- Nothing universal; each forecast needs evaluation on its own terms
- Every future prediction will fail
- Progress has stopped permanently
- The forecast was never made