On March 3rd, I was invited to speak on Pro TV's iLikeIT about a deceptively simple question that exposed the blind spots of today's AI assistants — and what it reveals about how we should interact with artificial intelligence.
The segment explored a question that went viral for stumping multiple AI assistants: "I want to wash my car. The car wash is 50 meters away. Should I walk or drive?" Every AI answered "walk, since it's only 50 meters away" — completely missing the obvious logic that you need the car at the car wash to wash it.
"When used correctly — with proper context and clarity — your chances improve significantly. Ask why. Demand proof. AI may then realize its own error. These models are powerful, but they process language statistically, not logically. They don't truly understand the situation — they predict the most likely next word."
The Trap: When Common Sense Beats AI
The question is a perfect example of how AI struggles with implied context. Any human immediately understands that the purpose of the trip is to wash the car — so you obviously need to drive it there. But AI assistants focused narrowly on the distance ("50 meters is short, so walk") without grasping the underlying goal. It's a reminder that these systems process language patterns, not real-world logic.
This kind of failure isn't rare. Ambiguous or contextually incomplete prompts regularly lead AI to confident but wrong answers — and most users don't think to question them.
Romanians and AI: Growing Use, Declining Trust
The discussion also touched on fresh data about AI adoption in Romania. Nearly 7 in 10 Romanians now use AI tools daily, especially in the 18–34 age group. But trust is shifting: the share of users who express high confidence in AI-generated information dropped from 65% in 2025 to 55% in 2026. People are learning — sometimes the hard way — that AI isn't infallible.
Better Questions, Better Answers
The core takeaway from the segment is practical: how you ask matters as much as what you ask. Vague prompts invite vague or wrong answers. Providing context, asking follow-up questions, and demanding reasoning from AI dramatically improves the quality of results. AI is a powerful tool, but it works best when guided by a critical, informed user.
As AI becomes woven into daily life for millions of Romanians, this kind of digital literacy — knowing how to work with AI, not just rely on it — is becoming an essential skill.
Whole news here -> https://stirileprotv.ro/stiri/ilikeit/capcana-care-a-incurcat-ai-ul-intrebarea-simpla-la-care-asistentii-inteligenti-au-raspuns-gresit.html