Writing04
Blog
01Apr 10, 202602Mar 20, 202603Feb 18, 202604Jan 24, 2026
How Context Shapes ChatGPT's Reasoning in Ambiguous Situations
Paired prompts — with context and without — put ChatGPT 5.3 Instant's moral reasoning under a microscope. Where it adapts, where it defaults, and why.
Do Large Language Models tell different people different things?
Five personas ask Claude the same question about antidepressants. A side-by-side look at how the answers shift with age, expertise and emotional state.
Exploring AI Untrustworthiness Through Card Games
Șeptică, a Romanian card game, as a controlled test of DeepSeek-V3.2 — where it skips game states, misreads the deck, and deals itself the better card.
Assessing the Reliability of Large Language Models in Reporting Sports Results
ChatGPT 5.2 gets the score right and the starting XI wrong. Checking a 2022 Romanian league match against Flashscore, and what that gap costs.