Be kind

A thing I like about small web and personal websites is that the people behind them are gentle and polite.

You could be a perfect stranger that get in touch with them for the first time, asking for feedback or just saying hello, and they will respond you gently, offering you help, giving you support and a connection, a real one, not like the one you get in many social networks.

In this context, being kind is like a superpower, because of its role of facilitator.

It helps removing walls and barriers, reducing the distance with other people, creating the right context for building. Building new relations, new opportunities and a sense of humanities.

Is not only a matter of netiquette, it seems to me more a kind of openness mood that reveals our need to socialise and considering people as such and not as “users”.

While writing about that, I can’t stop thinking that kindness it’s codified into human genetic.

I also think that it’s up to us to cultivate it and if / when activate it, based on our unique sensibilities, but above all, that fear is the greatest limiter of kindness.

All very profound, but since every thought of mine ends up connecting to the technological-digital sphere, what about machines and robots? Will we be able to give them this humans trait? Will we would even desire such a thing?

In theese days, while pondering on this topic (my writing process takes days), I saw a post on Linkedin talking about this, more or less.

Unfortunately I can’t remember where, it was form a page, not from a person, but can’t say which page.

It said that LLMs resent of the tone of the prompt, and this is known, but also that using kindness in a prompt tends to generate better responses, just as politeness does in human relations.

Searched for it on Google I found several articles about this topic, but with a different study as reference, and often without references at all. I will update the post If I’ll find the original post again.

For now I share this:
LLMs mirror certain human communication traits, which means politeness toward chatbots tends to generate better responses, just as politeness does in human conversation.

And this paper.

After that I asked to ChatGPT for help:

Hi, I need some help in better understanding something: I read an article that talks about some research regarding the fact that, using a kind and respectful tone in a prompt made toward an LLM like you, allows for better output, with less fabrication and higher quality in the response. What do you think this might depend on? Thank you

Its response:

Why a Kind and Respectful Tone Matters in Prompts

Politeness Patterns: Language models are trained on large datasets where polite language often appears in high-quality content. This association may lead to better responses.
Social Norms: Respectful language may mimic positive social interactions, influencing the model to generate more accurate and thoughtful replies.

Clear Communication: Polite prompts are often more structured and clear, reducing ambiguity and helping the model understand the request better.
Less Ambiguity: Kindness often correlates with precision in language, leading to more relevant and concise answers.

Attention Focus: Respectful prompts might simplify the language processing, allowing the model to focus better on the core of the question.
Reduced Noise: Hostile or unclear language might introduce "noise," potentially skewing the model's output.

Pattern Recognition: The model may associate polite language with informative, factual content, improving output quality.
Positive Associations: Rude tones might trigger associations with less accurate or biased content, decreasing response quality.

Social Simulation: The model mimics human interaction, where kindness often yields better cooperation and understanding.

Conclusion Using a kind and respectful tone when interacting with language models can lead to better, more accurate responses by leveraging the model's training patterns, reducing ambiguity, and fostering a positive simulation of human interaction.

I can't say how correct the output of my prompt is, but It seemed funny to ask.

Things are running fast with AI and I don't know how it will evolve, but I think we should train it with the best part of us.

In the future I'd rather meet a kind terminator than not.