ChatGPT trounces humans in entrance exams for top Japan university, study finds 100%

By Yukana Inoue0%

4/30/2026, 7:38:00 AM

BS Summary: This article contains 20 faulty reasoning types, including Framing Effect, Halo Effect, and Representativeness Heuristic, with Appeal to Authority as the most egregious example at 57.9% saturation with 81 hits. Analysis detected 857 faulty-reasoning hits from 140 analyzed words, generating a BS Score of 100% and a BS Rank of 100% (29 of 16,813 articles). This article is worse (more manipulative) than 99.80% of the article peer group.

If an artificial intelligence model such as ChatGPT had taken the entrance exams for Japan’s top university in 2026, it would have been assessed as top of the class and admitted for scoring higher than any human test takers, a study by AI startup LifePrompt has found. 
The research used three major AI models  ChatGPT 5.2 Thinking by OpenAI, Gemini 3 Pro Preview by Google and Claude Opus 4.5 by Anthropic  and had them take the actual entrance exam used by the University of Tokyo in February 2026 to assess candidates for courses set to start in April. 
The university’s category 3 science exam, often taken by those who want to enter the institution’s medical school, is considered the most difficult exam to pass in Japan. 
Confirmation Bias
33.6%
Anchoring Bias
0%
Availability Heuristic
20%
Representativeness Heuristic
37.9%
Hindsight Bias
0%
Overconfidence Bias
33.6%
Framing Effect
42.1%
Loss Aversion
0%
Status Quo Bias
20%
Sunk Cost Effect
0%
Optimism Bias
33.6%
Pessimism Bias
0%
Negativity Bias
28.6%
Self-Serving Bias
0%
Fundamental Attribution Error
0%
Actor-Observer Bias
0%
In-Group Bias
0%
Out-Group Homogeneity Bias
0%
Halo Effect
42.1%
Horn Effect
0%
Dunning-Kruger Effect
0%
Recency Bias
0%
Primacy Effect
0%
Blind-Spot Bias
0%
Ad Hominem
0%
Straw Man
0%
Appeal to Authority
57.9%
False Dilemma
0%
Slippery Slope
0%
Circular Reasoning
0%
Hasty Generalization
8.6%
Red Herring
0%
Bandwagon
0%
Appeal to Emotion
0%
Begging the Question
33.6%
Post Hoc (False Cause)
33.6%
Tu Quoque
0%
Burden of Proof
37.9%
Appeal to Nature
0%
Composition/Division
0%
Anecdotal
20%
No True Scotsman
0%
Ambiguity (Equivocation)
33.6%
Gambler’s Fallacy
0%
Middle Ground
0%
Personal Incredulity
0%
Special Pleading
0%
Genetic Fallacy
0%
Unattributed Quote
20%
Quote-first Misdirection
0%
Biased Writer Voice
8.6%
Indoctrination
33.6%
Politically Left Leaning Bias
0%
Politically Right Leaning Bias
0%
Attempt to Sell a Product or Service
33.6%

140 words analyzed.

Analysis

Hover over highlighted words in the article to view the associated bias or fallacy analysis.