ChatGPT is a chatbot that works thanks to an artificial intelligence model. So ChatGPT has no feelings. I am clarifying this because, having previously discussed information of the type covered in this report, this text has sparked several conversations in which the above has been repeated to me, both actively and passively, even though it is something that I obviously I am more than aware of this and I understand that every one of our readers is at the same point.

If we are kinder in our questions on ChatGPT, the chatbot will not rate them more positively and as a response based on gratitude for our kindness, will give us a more complete and accurate answer, etc. Now, even if this is not the reason, we already talked about this a few months ago good manners and courtesy According to Kurtis Beavers, director of the Microsoft Copilot development team, using Bing can lead to better responses.

If a chatbot is designed to mimic human conversation and is trained with datasets that reproduce said model thousands of times, it makes sense to think that once functional, it reproduces the specified communication pattern. The ability to infer an answer based on a question, training and modifying a model can be greatly affected by this aspect, because at the end of the day we are talking about human-made models, trained on human-generated data. and which try to reproduce human patterns.

Now we find a study published on arXiv that states this showing our emotions when using chatbots like ChatGPT can give us better responses. Several large-scale language models (LLM) were used for this study, and according to the results, it seems that the good results of using this technique in our queries apply to all of them, so we can understand that it is common to all chatbots.

Thus, according to this study, when conducting a consultation that integrates samples of our emotions about this (this would make me very happy, this is very important to me, etc.) or motivational (I know you can do this phenomenally, I am proud of you) and the like) led to overall performance improvement 8% in task results such as “Rephrase the sentence in formal language” and “Find the common characteristic of these objects” aa 10.9% on knowledge questions.

Again, I know AIs don’t have feelings, but we’ve already seen two testimonials about the positive, for our own interests, that we can use our own when using these tools.