Being nice to your chatbot turns out to be more than just digital decorum. According to a recent research, users who mind their manners with artificial intelligence (AI) assistants get faster, more accurate help, while their ruder counterparts face frustration.
For businesses investing heavily in AI customer service, the findings suggest a critical shift in strategy: promoting courteous customer-bot interactions could boost satisfaction rates and sales while reducing the need for human intervention, potentially saving companies millions in support costs.
“When customers interact politely with AI assistants, they’re unknowingly activating more thorough and careful response patterns, similar to how “think step by step” prompting improves problem-solving accuracy,” Dev Nag, CEO of QueryPal, a support automation company, told PYMNTS. “This isn’t just about being nice — it’s about triggering more reliable cognitive patterns in the AI.”
Large language models (LLMs) show significant performance variations based on how politely they’re asked to complete tasks, according to a study from researchers at Waseda University and RIKEN AIP. The team found that while rudeness generally leads to poor results across English, Chinese and Japanese prompts, excessive politeness doesn’t necessarily improve performance, with optimal levels varying by language and cultural context.
The study, which evaluated tasks including summarization, language understanding and bias detection, revealed that models trained primarily in one language are particularly sensitive to politeness in that language. Researchers also discovered that reinforcement learning from human feedback (RLHF) and supervised fine-tuning influence how models respond to different politeness levels. The findings suggest that developers should consider cultural nuances when training and deploying LLMs.
“For businesses, this creates a powerful opportunity to improve both AI performance and customer satisfaction simultaneously,” Nag said. “When companies encourage polite interaction with their AI systems, they’re not just promoting better social norms — they’re actually optimizing their AI’s performance in real time. The data shows that polite queries tend to receive more detailed, accurate and helpful responses, leading to higher resolution rates and customer satisfaction. Think of it as similar to how a skilled customer service manager might coach their team to maintain professionalism even with difficult customers — it creates a virtuous cycle where better interaction patterns lead to better outcomes.”
Nag said that from a customer experience perspective, encouraging polite interaction subtly guides customers toward interactions that will serve them better while maintaining their agency.
“Companies like Apple have already seen that when customers engage more respectfully with their AI assistants, they tend to provide clearer information about their needs and are more receptive to the AI’s suggestions and solutions,” he said. “This creates a win-win situation where the technical limitations of AI systems are partially mitigated by better interaction patterns, while customers receive more satisfying and effective service. The key is to make this feel natural and beneficial rather than forced or artificial.”
Studies show that a polite approach to chatbots isn’t just good manners — it actually improves the interaction. Research from Stanford reveals that polite users often experience fewer misunderstandings, as chatbots respond more accurately to courteous language. Meanwhile, a University of Cambridge study found that respectful phrasing prompts better responses, making exchanges smoother and more efficient.
There’s more at stake than just getting answers, though. Experts say that politeness to digital agents fosters empathy and reinforces positive communication habits, impacting how we interact offline. In workplaces and schools, modeling politeness with AI helps create a respectful digital culture that extends into real-world conversations.
“What these studies seem to indicate is that when it comes to chatbots, as with most anything else, a user-friendly interaction is most easily attained when the user plays nicely,” Zack Kim, founder and CEO of mental health platform Cartha AI, told PYMNTS.
For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.
We’re always on the lookout for opportunities to partner with innovators and disruptors.
Learn More