Should you be nice and polite to your chatbot, or treat it like a slave?

“Please” and “thank you” are the magic words we are taught to say as children. Also to say hello before asking someone something or say goodbye before leaving. But, Does this “human” protocol also apply to the conversation with a chatbot of Artificial intelligence ?

If you use ChatGPT , Gemini or another chatbot, You might be wondering how nice you should be to AI. Or if, on the contrary, you were to speak to him more directly and only give him orders, as if he were a servant or a slave.

This ethical debate is generated today in conversations with family and friends, and there are those who prefer to be kind to their chatbot, thank them and even give them compliments.

“You are a rock star,” Choudhary Sahni, an American founder of an AI company, told ChatGPT after giving him the precise answer to his questions.

Usually the man thanks the AI ​​a lot, “especially if I think I was rude to him before,” he said. The Wall Street Journal .

On the other hand, his wife Ridhi Sahni, an employee in a technology company, is not interested in being cordial with the robot.

“If I use a microwave, I don’t say something like ‘Dear LG, please reheat my food.’ I just press a button and go on with my day. “I think it’s purely transactional,” he said.

But, Is there any benefit to speaking nicely to an AI chatbot? Or on the contrary, will we get better answers if we are direct and get to the point, without being polite?

The development of artificial intelligence has progressed rapidly over the past few years. NOW, We can talk to robots, as if we were talking to another human.

Additionally, chatbots like ChatGPT or Gemini They are programmed to be cordial in their dealings with humans.

However, not everyone reciprocates the kindness. Some don’t say hello, ask please, or thank when the AI ​​gives them feedback on the questions they ask.

But, Is it wrong to speak harshly to them?

There is another group of people who feel bad for not saying hello, asking please, and thanking the searches and answers that AI does for them.

This usually happens because Conversations with a chatbot are so much like a conversation with a human, that there are those who can’t help but be cordial, as they would be with their peers. And they feel that on the other side, there is also a part of humanity.

According to a recent survey by Speaker Research near the 48% of respondents said it was important to be kind to AI, while the 27% said it was okay to be rude to robots.

And although everyone has the freedom to speak as they wish to the AI, apparently, Being nice to a chatbot could pay off in terms of responses.

According to a Japanese study, Being nice to chatbots, like ChatGPT, can have benefits: responses can be more accurate and complete.

Instead, If your instructions to the AI ​​are crude, it “may lead to increased bias, incorrect responses, or response rejection.”

This was demonstrated by the chief scientist of Microsoft Jaime Teevan. In conversation with WSJ, the expert said that “If you speak rudely to the model, you can expect her to be difficult with you as well.”

And, at least, its AI is designed not to react well to bad behavior, because The idea is that he most closely resembles a human.

“Just like humans, AI cannot always be the best person” they said from Microsoft on their blog.

Furthermore, experts in this technology have stated that, to verify this function, You can add specific phrases to make the AI ​​produce better answers. For example, you can tell him “take a deep breath before answering” to make him think a little better.

Even an experiment carried out by the same medium revealed that By offering ChatGPT “hints” for answering questions, it has improved the type of answers provided.

When told there would be a $200 tip, responses were 11% longer. In contrast, when a $20 tip was mentioned, the response was 6% longer than if it offered nothing.

The engineers are convinced that Generative AI has a “praise bias” and responds well to positive affirmations and rewards.

For example, Alana O’Grady, an executive at a California technology company, said: begins his questions to ChatGPT with “Could you please…?” and ends with “Good job!”

The expert is convinced that AI helps him better answer his questions by being kind.

“The litmus test today of how good a person you are is whether you are nice to a waiter. “In the future, it will be how nice you are to your AI partner” he added.

AI is increasingly endowed with human capabilities, so much so that many people They are turning to chatbots to help them treat others with kindness.

Software developer Laszlo Deak said WSJ that He usually uses the chatbot to vent when he has a problem at work, but also to suggest ideas on how to express this frustration to his boss or colleagues.

For example, he asks ChatGPT how to constructively tell another team of workers that their product is bad, because he didn’t know how to phrase it in kind words to make the point without crossing the line of respect.

“When you’re in the moment and you’re angry, it takes extra effort to reframe it. Reading the ChatGPT version helps me calm down, “ said the expert.

Something similar was discovered by Mazen Lahham, a technology manager in Dubai: he assured that uses AI in its business to calm angry and aggressive customers, a task that was more difficult for a human worker to accomplish successfully.

“The AI ​​has learned to absorb and respond in a calm and professional manner, which can be very difficult for a human,” he declared on his LinkedIn account.