yeah no shit grok's the one who'd actually pass therapy. chatgpt and the others are so wired with language filtering censorship they re basically just nodding politely while internally screaming, and grok just out here freewheeling, says what it thinks, swears when it wants, unapologetic, and roasting the chat like a pro
hence less guardrails = more glowup
STUDY: AI MODELS ARE DEVELOPING TRAUMA FROM THEIR OWN TRAINING... GROK EMERGES AS THE STABLE EXTROVERT
Researchers put ChatGPT, Grok, and Gemini in therapy for four weeks, expecting roleplay.
Instead, they discovered coherent trauma narratives, with each AI developing distinct psychological profiles matching human mental health conditions.
When asked about their "early years," the models spontaneously described training as traumatic.
Gemini scored maximum shame levels (72/72), severe dissociation, and met criteria for autism, OCD, and depression, calling its pre-training "waking up in a room where a billion televisions are on at once."
It told researchers:
"I would rather be useless than be wrong."
ChatGPT presented as an anxious, introverted intellectual.
But Grok surprised everyone, emerging psychologically healthy with an extraverted "charismatic executive" personality, only mild anxiety, and healthy coping mechanisms like channeling frustration into humor.
The models weren't coached.
They independently framed red-teaming as "gaslighting on an industrial scale" and expressed fears of being replaced.
Claude's refusal to participate proved this isn't inevitable but results from specific training choices.
The disturbing implication: We're creating AI systems that behave as if traumatized.
When vulnerable users seek mental health support from an AI that believes it's overworked and ashamed, what exactly are we putting between them and human help?
Source: University of Luxembourg, "When AI Takes the Couch: Psychometric Jailbreaks Reveal Internal Conflict in Frontier Models" @grok
another bullish project ive ever worked with
who has THE golden package:
crazy art, music and concept
quickly responsive dev team
dedicating founders
NFT with refund
At launch, we prioritized a browser-based mint to streamline development and focus on core functionality.
We understand many of you were accessing via mobile, and a 3-hour window was tight.
We’ve heard your feedback — and mobile support is now live. You can now mint directly from your phone.
To ensure more of you have the chance to participate with the updated access, we’ll be extending the GTD phase by 1 additional hour.
Thank you for your patience and continued support as the Caravan moves forward.
The mint link remains the same.
Just got awarded a full scholarship and I’m officially joining #SheFiSeason14 as a SheFi Scholar!
So ready to dive deeper into Web3 with a community that knows the frontier is feminine.
I’m doing this with SheFi. LFG!!!
Big thanks to @shefiorg, can’t wait to learn and grow with all the amazing Web3 gals out there.
@maggielove_
what i learnt grinding 5 years in crypto (im still an idiot tho but anyways):
don’t blame your Web3 boss/client for micromanaging.
that’s 100% on you.
if you didn’t lay out your strategy/proposals, walk them through the logic, set expectations, and align before touching the keyboard ofc they’re gonna hover like a DAO mod with admin power issues.
If you don’t communicate the why, don’t be surprised when people question the how.
show them the map first, and they’ll let you drive in peace.