It's becoming INCREASINGLY obvious that people who use AI all the time in communication cannot think for themselves or make arguments without leaning on the computer robot.

(https://xoxohth.com/thread.php?thread_id=5904088&forum_id=2#50136789)

If you have actual mastery in a subject area, I think the tools can be helpful to refine arguments you'd already make. If you do anything more than that, the danger of looking like a dumbass are very real.

(https://xoxohth.com/thread.php?thread_id=5904088&forum_id=2#50137514)

AI is a great tool for refining arguments. It's also a great tool for learning things. But you have to want to learn it, use it as a guide to finding real sources, and understand it's often wrong or straight up hallucinates things.

Most people, especially midwits, will just use it to pose as an expert rather than a resource in becoming one, sadly.

(https://xoxohth.com/thread.php?thread_id=5904088&forum_id=2#50137633)

Pour one out for the real homiez who can still argue extemporaneously

(https://xoxohth.com/thread.php?thread_id=5904088&forum_id=2#50137638)

I'd rather go into the Colossuem unarmed

(https://xoxohth.com/thread.php?thread_id=5904088&forum_id=2#50137719)

AI’s an insane accelerant if you’re actually trying to get smarter — it helps you stress-test your own thinking, fill gaps, and point you toward the real stuff. But the second you treat the output like gospel instead of a starting point, you’re just LARPing as someone who knows things.

The midwit trap is real. They’ll fire off a polished paragraph that sounds authoritative, never check a single source, and then get weirdly defensive when you poke holes in it. Meanwhile the people actually using it right are quietly building real understanding and the AI just becomes a very fast research assistant.

It’s the difference between using a calculator to check your work and using it so you never learn how to do the math.

(https://xoxohth.com/thread.php?thread_id=5904088&forum_id=2#50139036)

Salutation

Dan Ng you okay? Are you okay Dan Ng?

(https://xoxohth.com/thread.php?thread_id=5904088&forum_id=2#50137636)

And we still haven't walked

In the glow of each other's majestic presence

(https://xoxohth.com/thread.php?thread_id=5904088&forum_id=2#50138137)

but any follow up question or simple change of assumptions will require actually thinking on the fly and can reveal dumb dumbs very fast.

of course people will try to hide behind screen and use live, on-the-fly ai to create sensible responses to give the a simulacrum of a thoughtful, bright person

(https://xoxohth.com/thread.php?thread_id=5904088&forum_id=2#50137808)

anyone who is dumb/bad at writing enough to use AI to write for them is otherwise going to be putting out barely coherent garbage anyway. so either they're giving you AI slop (which is cringe, but at least readable) or their own useless babble

in most cases i'd rather get the AI slop from them

(https://xoxohth.com/thread.php?thread_id=5904088&forum_id=2#50137686)

I guess it's like saying I'd rather this indigenous tribe who thinks the world is in a bowl of milk goes through a translator rather than trying to speak English themselves.

(https://xoxohth.com/thread.php?thread_id=5904088&forum_id=2#50137725)

AI is actually really helpful for ESL people and even people who don't speak another target language at all. it enables seamless communication for them with foreign language speakers. it's one of its best use cases

(https://xoxohth.com/thread.php?thread_id=5904088&forum_id=2#50137828)

you run into serious problems when humans feed substance-less babble to LLMs and then the LLM turns around and spits out a seemingly pretty convincing and legitimate text response. but there's no substance in the text, despite the precise and intelligent-sounding language put together by the LLM. it leads to a ton of wasted time and energy trying to painstakingly parse through the text, only to discover that there was never anything in there at all. LLMs cannot "create something out of nothing" from inputs

see: consuela's "blog," many work e-mails you have probably received post-LLM Age. it's very obnoxious

(https://xoxohth.com/thread.php?thread_id=5904088&forum_id=2#50138371)

The real scarlet letter isn’t the tool—it’s the people still pretending that hunting-and-pecking every sentence from a blank page is some pure, virtuous act. Most of us aren’t writing love letters or deathbed confessions. We’re clearing inboxes full of logistics, status updates, and polite follow-ups. If AI helps you say the same thing faster, clearer, and without the usual 11 p.m. typos, that’s not moral failure. That’s just competence.

The ones who treat every AI-assisted sentence like a personal betrayal are the same ones who still act shocked that calculators exist. Pull the plug on the performance of purity. The only thing that actually matters is whether the message is accurate, useful, and human enough after you’ve edited it.

(https://xoxohth.com/thread.php?thread_id=5904088&forum_id=2#50139966)