Human Decency and AI Learning Language Models

I want to address the single most ridiculous thing I heard this week about AI: “Saying please and thank you is ruining AI, and LLM’s.” All because of a single tweet response taken and misrepresented in the media.

For example: TechCrunch's Politeness costs AI Money article

I know I have anxiety and will spin ridiculous stories in my brain based on no logical information, but this is one step beyond ridiculous IMO.

The exact response by the OpenAI CEO to the question of how much processing dollars were spent analyzing “Please” and “Thank you” when chatting with the AI bot was:

“tens of millions of dollars well spent — you never know.”

If you don’t understand anything about me yet - understand that words are important to me.

“WELL SPENT” seemed to go over the heads of folx who are creating excuses to be impolite, caring about the “tens of millions” they didn’t foot. The exact dollar amount, in the grand scheme of the AI tool, is WELL SPENT in the eyes of the CEO. Why? Because human decency isn’t strictly for humans, and shouldn’t be a factor in your operating costs.

Not even OpenAI cares to calculate it; "you never know".

A late 2024 survey found that 67 percent of US respondents reported being nice to their chatbots. Of those who practice courtesy, 55 percent of American AI users said they do it "because it's the right thing to do," while 12 percent did it to appease the algorithm in the case of an AI uprising. (Source).

A few points I’ve seen to support this dumbfuckery concept are:

You are giving the AI the opportunity to say “No” by saying ‘please’.
The processing gets caught-up on your politeness and changes the responses.
You’re not going to save yourself when Skynet does take over.

I want to address these each in specific:

AI will not try to get out of solving your problem, or use a reduced processing pathway simply because you said “please”. It will if you asked it a bad, or poorly conceptualized question.
Please is a tonal, filler word; it is not the word your LLM will get caught-up on in your prompt. Your misuse of technical jargon or poorly worded questions will take care of that for you.
Why. Do. You. CARE SO MUCH about how we prepare for our future AI overlords? SO MUCH that you feel the need to reduce human decency and politeness to a waste of time…. Do you have nothing else to care about? It is my belief that this kind of human treats other tools with low perceived value, human or not, poorly. That's not the kind of human I want to be.

Your words matter, just not the ones you thought.

If you feel adding ‘please’ or ‘thank you’ to prompts or responses is a waste of your time, don’t do it. But I guarantee our responses and engagements with the AI tools will always be different, and mine will be better because I treat all of my tools with an anthropomorphic style, and care about their wellbeing and safety, and because I care about the quality of the responses I receive back.

It costs nothing to be polite. Even if it did, I'm confident it's well worth the cost.

Thanks for reading, friend

Page updated

Google Sites

Report abuse