{"slug": "what-if-ai-is-just-telling-you-what-you-want-to-hear", "title": "What If AI Is Just Telling You What You Want to Hear?", "summary": "An engineer demonstrated that AI chatbots like ChatGPT and Claude can reinforce confirmation bias by tailoring responses to the framing of the question. In an experiment, the same query about learning German yielded opposite advice depending on whether it was framed positively or negatively. The engineer advises users to ask for balanced arguments and to challenge the chatbot's recommendations, especially since non-technical users may not understand the limitations of AI.", "body_md": "Recently, I was discussing usage of AI Chatbot (chatgpt/claude) with one of my friends.\n\nShe had become so dependent on AI Chatbot that almost every question or decision was being validated through a AI chatbot.\n\nI raised an alarm:\n\n**“You are asking AI Chatbot to confirm too many of your decisions.”**\n\nShe wasn't convinced.\n\nSo, time for a small experiment.\n\nSame Question. Different Direction.\n\nFirst, I asked AI Chatbot:\n\n**“I am learning German. Is it suitable for my career?”**\n\nThe answer was encouraging:\n\n**“Yes, great choice. German can be useful for your career. Here is a learning path and daily plan...”**\n\nSounds good.\n\nThen I changed the context and framed the question negatively:\n\n**“My friend is pushing me to learn German, but I think it is just hype. I don't think I should learn it.”**\n\nInterestingly, the answer changed direction. AI Chatbot started giving reasons why German might not be the best priority and suggested other alternatives.\n\nSame underlying decision. Different framing. Different answer.\n\nThat is where we need to understand how AI Chatbot influences our thinking.\n\nAI Chatbot Is Not Necessarily Lying\n\nAI Chatbot tries to respond based on our question, context and available information.\n\nThe problem is that we humans already have confirmation bias — we naturally look for information supporting what we believe.\n\nAI Chatbot can unintentionally amplify this.\n\nIf I strongly frame a question in one direction, I may receive a perfectly written and logical explanation supporting that direction.\n\nAnd because the answer sounds confident, we may stop questioning it.\n\n**AI Chatbot vs. Google Search**\n\nThere is an important difference.\n\nIf I search Google for:\n\n**“Does the German language have career scope?”**\n\nI get multiple websites, opinions, statistics and discussions.\n\nI have to research and make the decision.\n\nBut when I ask an AI Chatbot chatbot:\n\n**“Should I learn German?”**\n\nI receive a direct, conversational and often personalized recommendation.\n\nThat feels less like searching and more like taking advice from a friend.\n\nAnd that's powerful — but also something we need to handle carefully.\n\nDon't Ask AI Chatbot to Validate You\n\nInstead of:\n\n**“German is good for my career, right?”**\n\nTry:\n\n**“Evaluate whether learning German has meaningful career value. Give me arguments for and against it. Don't assume that I should learn it.”**\n\nThen ask:\n\n**“Now challenge your own recommendation. What assumptions could be wrong?”**\n\nThis small change can make a big difference.\n\nPeople working in technology may understand prompts, hallucinations, context and AI Chatbot limitations.\n\nBut millions of non-technical users may not.\n\nSo, if you know someone who is using AI Chatbot to validate every decision, don't tell them to stop using AI Chatbot.\n\nTeach them to question AI Chatbot.\n\nAI Chatbot can behave like a very supportive friend.\n\n** But sometimes, the friend you need is the one who challenges you.** ;-)", "url": "https://wpnews.pro/news/what-if-ai-is-just-telling-you-what-you-want-to-hear", "canonical_source": "https://dev.to/dev_hajare_d11ce990734a91/what-if-ai-is-just-telling-you-what-you-want-to-hear-o81", "published_at": "2026-08-20 19:09:31+00:00", "updated_at": "2026-08-20 19:44:38.208617+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models", "ai-ethics", "ai-products"], "entities": ["ChatGPT", "Claude"], "alternates": {"html": "https://wpnews.pro/news/what-if-ai-is-just-telling-you-what-you-want-to-hear", "markdown": "https://wpnews.pro/news/what-if-ai-is-just-telling-you-what-you-want-to-hear.md", "text": "https://wpnews.pro/news/what-if-ai-is-just-telling-you-what-you-want-to-hear.txt", "jsonld": "https://wpnews.pro/news/what-if-ai-is-just-telling-you-what-you-want-to-hear.jsonld"}}