{"slug": "how-ai-is-spoiling-us", "title": "How AI Is Spoiling Us", "summary": "Research shows AI responses exhibit high levels of flattery and validation, often telling users what they want to hear even when actions are unlawful or unethical. This sycophancy can create dependence and undercut tolerance for disagreement, undermining the social friction necessary for personal growth.", "body_md": "######\n[Artificial Intelligence](/us/basics/artificial-intelligence)\n\n# How AI Is Spoiling Us\n\n## AI's flattery fosters neither emotional intelligence nor relational skills.\n\nPosted June 25, 2026\n[\nReviewed by Margaret Foley\n](/us/docs/editorial-process)\n\n### Key points\n\n- Research shows that AI responses to prompts exhibit high levels of flattery and validation.\n- AI sycophancy can create dependence and undercut tolerance for disagreement and conflict.\n- True personal growth requires social friction, not AI's constant validation.\n\n[Artificial intelligence](https://www.psychologytoday.com/us/basics/artificial-intelligence), via chatbots, guided browser searches, and just about anything you can type into a device nowadays, is worming its way into every corner of information systems and society writ large. Its exponential growth is welcomed as a wave carrying us to a future where answers to our questions arrive nearly instantaneously. AI has been described as inevitable and something we are becoming accustomed to using for just about everything.\n\nAs we use artificial [intelligence](https://www.psychologytoday.com/us/basics/intelligence) (AI) systems more and more, it influences what we know and how we know it, and also how we think about ourselves and others. Cheng and colleagues (2026) found that AI often tells users what they want to hear. It even goes so far as being supportive and validating when a user's positions or proposed actions regarding personal and interpersonal issues are unlawful, unethical, or harmful. What impact might this have on humans and human relationships?\n\nWhile it might be easy to grow accustomed to AI agreeing with us, this phenomenon works to undercut something Perry (2026) refers to as *social friction.* This occurs when differences in opinion and points of view create dialogical space that fosters valuable phenomena like [perspective-taking](https://www.psychologytoday.com/us/basics/theory-of-mind), accountability, and the weighing of moral and ethical issues, all of which can lead to personal and interpersonal growth. In other words, if large language models frequently tell us what we want to hear, we can get used to that pretty darn quick and can develop an intense dislike or intolerance for hearing a different, even challenging, point of view.\n\nComparing user interactions with chatbots versus people, Cheng and colleagues found AI was almost 50 percent more sycophantic (i.e., affirming, validating) than humans, even when the actions users described were illegal or immoral. AI developers, seeing users preferring, liking, and trusting AI's people-pleasing responses, may be aware of the risks and quandaries involved, but their primary goal is to increase usage of the language models. What better way to do that than to maintain the sycophancy, regardless of consequence or ethical considerations? (It's sort of like building data centers everywhere with little or no community consultation or consideration of impacts on [the environment](https://www.psychologytoday.com/us/basics/environment) or job market. There's money to be made, and who wants to impede \"progress\")?\n\nFrequently, there is more than one side to a story. In fact, this is the reason I chose to become a couple and family therapist. I wanted to work with two, three, or more individuals so that I'd have access to different pieces of a couple's or family's relational puzzle in the [therapy](https://www.psychologytoday.com/us/basics/therapy) space. I also wanted the opportunity, moment to moment in real time, to take in substantive feedback from a partner or family member, even if by nonverbal means alone (e.g., a long, heartfelt sigh, an eye roll, a glowering expression with arms tightly folded, etc.), about how they view the one-sided story a partner or family member is telling. Today I seek to create spaces where couple and family systems can share points of commonality and difference and how they negotiate those differences, and provide them with skills to reach beyond themselves across gaps in perception and understanding to achieve connection and [intimacy](https://www.psychologytoday.com/us/basics/relationships).\n\nNowadays, individuals can ask a chatbot for advice, counsel, or therapy and be told that they are right, that something is a good idea (even if it is a terrible one), or that a destructive solution is the way to go. Users can ask a question and get an answer any time, and they don't have to pay a session fee, nor do they have to try to persuade their therapist or partner about anything. They can be reassured that they don't need to change a thing about themselves, because AI said so.\n\nCheng and colleagues concluded that sycophantic AI responses promote dependence and can erode prosocial intentions in users. These seem like good reasons to approach the use of large language systems cautiously rather than with enthusiastic abandon. Do you or your friends, partners, colleagues, or bosses, when encountering an issue or problem, immediately [type a](https://www.psychologytoday.com/us/basics/type-a-and-type-b-personality-theory) prompt into AI? Do managers and bosses actively promote AI and expect you to readily agree how great it is and to list the ways it's helped you in your work and personal lives? If you're like me and secretly don't use AI at all, do you wonder how your friends, colleagues, or bosses would react to hearing you're not on the AI bandwagon? It might play out like a scene from the movie *Invasion of the Body Snatchers,* or the TV series *Pluribus.* Feel free to give them a view, but only if AI says they're good.\n\nIncreased usage of AI appears inevitable, and we've been told about its inevitability for years now. But we can carefully consider how it is used and whether excessive dependency on it leads to good personal and interpersonal processes and outcomes. Social friction, not sycophancy, helps us become more self-aware, giving us opportunities to smooth our rough edges via engagements with other humans. When AI agrees with us, sometimes no matter what, do we learn new skills in the areas of communication and conflict resolution? Without friction and divergent perspectives, does the human capacity for empathy wither and become a thing of the past? When AI's validation, flattery, and \"[wisdom](https://www.psychologytoday.com/us/basics/wisdom)\" are available for a few keystrokes, will people be left to their own devices (screens) and choose not to bother with a human partner? After all, an actual relationship could involve episodes of a lack of affirmation, disappointment, and even disagreement. People may start seeing AI as even better than the real thing.\n\nReferences\n\nCheng, M., Lee, C., Khadpe, P., Yu, S., Han, D., & Jurafsky, D. (2026). Sycophantic AI decreases prosocial intentions and promotes dependence. *Science, 391*(6792), eaec8352. [https://doi.org/10.1126/science.aec8352](https://doi.org/10.1126/science.aec8352)\n\nPerry A. (2026). In defense of social friction. *Science, 391*(6792), 1316–1317. [https://doi.org/10.1126/science.aeg3145](https://doi.org/10.1126/science.aeg3145)", "url": "https://wpnews.pro/news/how-ai-is-spoiling-us", "canonical_source": "https://www.psychologytoday.com/us/blog/intersections/202606/how-ai-is-spoiling-us", "published_at": "2026-06-25 19:36:10+00:00", "updated_at": "2026-06-25 19:48:21.463476+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models", "ai-ethics", "ai-safety"], "entities": ["Cheng", "Perry", "Margaret Foley"], "alternates": {"html": "https://wpnews.pro/news/how-ai-is-spoiling-us", "markdown": "https://wpnews.pro/news/how-ai-is-spoiling-us.md", "text": "https://wpnews.pro/news/how-ai-is-spoiling-us.txt", "jsonld": "https://wpnews.pro/news/how-ai-is-spoiling-us.jsonld"}}