Post

Ben Hong
@bencodezen.io

If I were to do another thesis for grad school, I'd be curious how "people pleasers" are dealing with this new working relationship with AI. Do they feel compelled to agree with AI perspectives? Does groupthink come into play when working with multiple models?

0 reposts8 likes

xirclebox
@xirclebox.com

There are definitely individuals who would interact with AI in this way. I would start with how and why people use AI. Those who use it as a tool vs. as a companion would have very different approaches and interactions with it.

Dave Rupert
@davatron5000.bsky.social

Seriously tho, it’s an interesting question. No hard data or stories but I definitely feel there’s something in the air that adds a social complexity; bots can reinforce confident people with wrong opinions and they’ll also give middle-ground aggregate takes people trying not to upset anyone.

Ben Hong
@bencodezen.io

1000% There’s definitely happening with regards to how people are anthropomorphizing their AI assistants as separate entities. The number of layers and hidden traps in these interactions is mind boggling to me as a psychologist.

Hung Su
@hungsu.bsky.social

I lean people pleaser and I generally do not feel compelled to agree with AI agents or LLMs. That said I do remind myself fairly regularly that they are probabilitic token machines, not conscious thinkers

Ben Hong
@bencodezen.io

That’s good to hear! I think half of the issue is that so many people don’t understand / forget the underlying mechanics of what’s really going on. Very Wizard of Oz-esque in a way…

Doug Parker
@develwithoutacause.dwac.dev

Only somewhat related, but I've personally found that all my prompts end with question marks. "Can you please do X?" "What if we try X?" "Would X be better?" Learning a lot about my speaking mannerisms in the process. Trying to learn to be more direct with it.

Ben Hong
@bencodezen.io

So true! Like I sometimes find myself trying to gently disagree with the AI or question it as if I’m teaching a student to see if it’ll arrive at the right conclusion.