IPA IP 2025 Question 85
A question IPA Japan published in 令和7年度 ITパスポート試験 公開問題 問85, translated into English. ITPEC is based on the same exams.
There is an AI system that solves problems and chooses the correct answer from an answer group. Each time this system goes through one learning process, its error rate decreases by 5% in the comparison before and after the learning process (it becomes 95% of the previous error rate). When the current correct answer rate is 30%, at least how many more learning processes are needed for the correct answer rate to exceed 35%?
Explanation
2 more learning processes — work with the error rate, which shrinks by a factor of 0.95 each time, and check when the correct answer rate passes 35%.
Correct answer rate 30% means error rate 70%. Each process multiplies the error rate by 0.95:
- •After 1 process: 70 × 0.95 = 66.5% error, so 33.5% correct, not yet over 35%
- •After 2 processes: 66.5 × 0.95 = 63.175% error, so 36.825% correct, over 35%
The correct answer rate exceeds 35% once the error rate falls below 65%, and that happens after the second process.
Answer (b)
Why not others:
- •(a) 1 — one process brings the correct answer rate only to 33.5%; even reading the 5% as points added to the correct answer rate gives exactly 35%, which does not exceed 35%
- •(c) 3 and (d) 4 — more than needed; the rate is already 36.825% after two processes
Key rule: A percentage decrease applies to the error rate, not to the correct answer rate: multiply the error rate by 0.95 each step, then subtract from 100%.
Basic Theory · AI-written, may contain errors
This question comes from the past questions IPA Japan publishes, translated into English by ITPEC Practice. ITPEC Practice is an independent study tool and is not affiliated with IPA or ITPEC. See the original paper (PDF) or Report an issue.