In the increasingly popular AI era, many people have the habit of communicating with chatbots like ChatGPT with polite phrases such as thank you, you can, or thank you. However, such politeness is not only unnecessary but can also reduce the accuracy of feedback and waste processing resources.
In fact, according to AI technology companies, users using polite phrases makes generative AI models such as ChatGPT, Gemini AI or other chatbots take more time and energy to process.
This not only extends response times but also costs AI companies millions of dollars in operating costs per year. The reason is that the system has to analyze unnecessary grammatical factors, making the results less accurate.
In addition to social phrases, some "requited" words are also listed as not to be used. For example, can, if you can, you can help often make AI misunderstand that users are looking for permission, or not clear about the purpose. This can lead to key feedback.
Instead, developers recommend that users should write reminders (prompts) directly and clearly such as: "write a doan", "explain the concept", " list the steps", ... Minimizing commands helps AI process quickly and give results close to the real needs of users.
Some other words to avoid include:
- Increased vague words such as many, recent, regular.
- words with no meaningful language value such as really, basically, only.
- Phrases that express a lack of confidence like excuses can make AI respond with unnecessary tone.
To maximize the effectiveness of AI like ChatGPT, users should be more straightforward in reminders. In the machine world, politeness does not mean efficiency and sometimes, talking like a machine is the smartest way to exploit technology.
