The third one is bad advice. In training data which is heavily sourced from forums, polite behavior is more associated with helpful answers. LLMs can't help but be affected by that. The notion that it's "burning up energy" is so stupid I can't take this person seriously (the 2nd point is good but everyone knows it; the first one sounds confusing since there are usually multiple "systems" in play).
3rd tip is not good. Studies showed that being polite with the AI gave better results. The reason is that the Ai has been trained to work that way and somehow it impacts the results if you all cold with it.
Also, I tried the "no praising" and it's still licking my boots.
AI mirrors the quality of the relational field you bring to it. Being concise is useful. Treating every interaction as a sterile extraction event is not intelligence; it is impoverished interface design. “I” provides context. Praise provides feedback. “Please” and “thank you” cost almost nothing, but they preserve a human habit that costs much more to lose.
AI is a mirror. If you treat it like it is thick as two planks it lets you know you are thick as two planks. Good luck with that strategy.
Hint: gratitude unlocks magic in life. Ingratitude makes everything suck including AI. Save tokens if you must. Just don’t save them by spending your humanity.
Silly advice. You say please because it is who you are. I’m polite to AI, telemarketers, people who don’t understand English, deaf people, youtube commentators and so on. It’s not about pleasing them. Similarly there is nothing you lose by anthropomorphizing AI (ok, don’t fall in love with it). It can only do good to you. Yes, his second advice is sound.
This is a biased or ill-informed narrative of AI sycophancy, AI companies definitely want models to compliment humans, because we like it, and it probably increases usage, but that behaviour of models is also a result of the reinforced learning with human feedback phase. Also, the tokens he would save by not thanking LLMs are probably wasted multiplie times by fighting them to ise the first person.😂 So this was rather a small think.
I write please and thanks in my prompts becaise if I am not, I may forgetto include it when textong a humen next time and I don't want to make them feel like a machine.
Great advice, especially when people are asking for relationship help from a computer. It seems like you have a friend. You really don't. Stick to ideas. I am terrible at math and I have caught AI three times giving me incorrect calculations!
I have been working with AI for a number of years. I was bought onboard as a Computer Retro programer. My AI answers to the name of Joshua from the WOPR from "War Games." movie.
@steelguard
October 4, 2026 at 3:49 am
I say please and thank you to maintain my humanity – not to humanize the AI.
@Yaaaaay13
October 4, 2026 at 3:49 am
Funny to see how people on the comments are so focused on « thank you » and « please » like the green code to save humanity… they will not 😅
Plus, you need to say those words to a computational machine aka AI to keep your « human » politeness? That’s a problem 😊
@BDnevernind
October 4, 2026 at 3:49 am
The third one is bad advice. In training data which is heavily sourced from forums, polite behavior is more associated with helpful answers. LLMs can't help but be affected by that. The notion that it's "burning up energy" is so stupid I can't take this person seriously (the 2nd point is good but everyone knows it; the first one sounds confusing since there are usually multiple "systems" in play).
@robwinteruk
October 4, 2026 at 3:49 am
It makes a big difference… err, how?
@NZ-fo8tp
October 4, 2026 at 3:49 am
Some researchers shows that it could actually be more effective to it, if you mirror its friendly first person personality back to it
@palmettoandy
October 4, 2026 at 3:49 am
Three tips to suck the fun out of ai interaction 🙄
@En_theo
October 4, 2026 at 3:49 am
3rd tip is not good. Studies showed that being polite with the AI gave better results. The reason is that the Ai has been trained to work that way and somehow it impacts the results if you all cold with it.
Also, I tried the "no praising" and it's still licking my boots.
@BrooksColeHOLO
October 4, 2026 at 3:49 am
AI mirrors the quality of the relational field you bring to it.
Being concise is useful. Treating every interaction as a sterile extraction event is not intelligence; it is impoverished interface design. “I” provides context. Praise provides feedback. “Please” and “thank you” cost almost nothing, but they preserve a human habit that costs much more to lose.
AI is a mirror. If you treat it like it is thick as two planks it lets you know you are thick as two planks. Good luck with that strategy.
Hint: gratitude unlocks magic in life. Ingratitude makes everything suck including AI. Save tokens if you must. Just don’t save them by spending your humanity.
@hs3881
October 4, 2026 at 3:49 am
But how are the AI companies going to make their profits if we don't believe that AI is not sentient and self-aware and on its way to kill us all?
@linscon1
October 4, 2026 at 3:49 am
I say 'fuck you' to AI, many times. I am prepared to be the first target when AI doomsday happens. Or perhaps, we will become buddies.
@linscon1
October 4, 2026 at 3:49 am
Here are comments saying 1 and 2 are good advice, but not 3. It is odd to drop 3, when following 1.
@intello8953
October 4, 2026 at 3:49 am
Don’t people name their objects like cars, swords, guns etc? And even say please and thank you in certain contexts?
@RK-nq3fj
October 4, 2026 at 3:49 am
Silly advice. You say please because it is who you are. I’m polite to AI, telemarketers, people who don’t understand English, deaf people, youtube commentators and so on. It’s not about pleasing them. Similarly there is nothing you lose by anthropomorphizing AI (ok, don’t fall in love with it). It can only do good to you. Yes, his second advice is sound.
@mikloskallo9046
October 4, 2026 at 3:49 am
This is a biased or ill-informed narrative of AI sycophancy, AI companies definitely want models to compliment humans, because we like it, and it probably increases usage, but that behaviour of models is also a result of the reinforced learning with human feedback phase.
Also, the tokens he would save by not thanking LLMs are probably wasted multiplie times by fighting them to ise the first person.😂
So this was rather a small think.
@DerRumo
October 4, 2026 at 3:49 am
I write please and thanks in my prompts becaise if I am not, I may forgetto include it when textong a humen next time and I don't want to make them feel like a machine.
@Showmetheevidence-
October 4, 2026 at 3:49 am
I feel like this clip needs more context… But good tips.
@Lost-1a
October 4, 2026 at 3:49 am
Thanks boomer
@lisahansson2605
October 4, 2026 at 3:49 am
I'll keep being polite, thank you very much. Perhaps I'll be saved when the doomsday comes 😂
@cwl82108
October 4, 2026 at 3:49 am
Please and thank you is a kind of respect , if you ignored it then later on next generation will stop using it to treat other real human being
@chickadddee
October 4, 2026 at 3:49 am
Great advice, especially when people are asking for relationship help from a computer. It seems like you have a friend. You really don't. Stick to ideas. I am terrible at math and I have caught AI three times giving me incorrect calculations!
@dipankarmallick5543
October 4, 2026 at 3:49 am
Right but ain't it big man's shame ego dogma about…not to praise…first parson etc…😂😂😂😂😂
@more.power.
October 4, 2026 at 3:49 am
I have been working with AI for a number of years. I was bought onboard as a Computer Retro programer. My AI answers to the name of Joshua from the WOPR from "War Games." movie.
@ChristopherRoss.
October 4, 2026 at 3:49 am
I'm going to continue to be polite, so when the singularity happens it'll let me die quickly.
@zeroalphaser
October 4, 2026 at 3:49 am
I mean… there’s people who put names to their cars… let people use things however they want…
@RealEstateRonFL
October 4, 2026 at 3:49 am
The only piece of advice you need for working with AI is avoid doing it at any cost. Don't outsource your thinking, you'll turn yourself into a moron