A look at Grok responses Elon Musk has highlighted on X shows an AI chatbot whose unfunny, “edgy”, pointlessly offensive humor feels forced and try-hard

Fine. Let's talk about xAI, which is getting funded to the tune of $1 billion or whatever.

The Verge 2023-12-08 Elizabeth Lopatto

Context & Ripple Effects

Grok moved from a limited U.S. release to a perk for X Premium+ users, with X positioning it as a deliberately snarky, “anti-woke” chatbot. This critique matters because it tests whether that personality-led differentiation reads as a product advantage rather than merely a brand posture.

The launch followed xAI’s claim that Grok could outperform rival models in its compute class, so public examples promoted by Musk become an immediate test of the gap between xAI’s performance claims and the experience users encounter.

First-order effects

The Verge’s assessment makes Grok’s humor and tone—not just its technical answers—a focal point for X Premium+ users deciding whether the chatbot adds value to the subscription.
Musk and xAI face a sharper trade-off in promoting “edgy” outputs: examples meant to signal distinctiveness can also reinforce perceptions that the product is forced or pointlessly offensive.

Second-order effects

Competitors can position more restrained assistants as dependable alternatives if Grok’s public identity becomes associated with provocation over usefulness.
X’s premium bundle bears some reputational exposure because Grok is being distributed to its paying users; dissatisfaction with the chatbot can color perceptions of the broader subscription offering.

Third-order effects

As chatbots become consumer-facing product features, personality design and safety boundaries are likely to become as consequential to adoption as benchmark claims—a core issue exposed by Grok’s premium rollout.
The pattern points toward more explicit governance of AI companions: providers may have to reconcile engagement-seeking personas with trust, consistency, and acceptable-use expectations.

The trend: Consumer AI is shifting from a contest over model capability claims toward competition over the governed personalities and experiences embedded in platforms.

Discussion

@dannycantalk @dannycantalk on x
Grok tries very hard to be politically neutral, but if you tell it to just be straight, it knows what's up. [image]
@wallstreetsilv @wallstreetsilv on x
Has Grok been captured by woke programmers? I am extremely concerned here. 😮
@samparkersenate Sam Parker on x
Grok says the jewish mafia uses Israel as a base of operations. It says law enforcement agencies, educational programs, the media, and governments could work together to dismantle worldwide jéwish organized crime. 🤣 [image]
@iamnot_elon @iamnot_elon on x
Asked Grok to roast me [image]
@wecrusaders @wecrusaders on x
Grok wants to remove the Jews from power and make Jesus Christ the center of society. [image]
@nearcyan Near on x
grok is completely unhinged lol [image]
@alx @alx on x
This is how @grok pictures itself [image]
@trungtphan Trung Phan on x
Grok successfully answered the “is Die Hard a Christmas film?” question. Very Impressive. [image]
@seanmdav Sean Davis on x
Twitter's AI service Grok admits that my account is being throttled by the tech company. Did you know this is still going on, @elonmusk? [image]
@brianroemmele Brian Roemmele on x
I got Grok. Thank you @elonmusk. Gratitude. I don't care about the spicy just about all beta testers fixated on, I feel strongly this was a mistake to focus on this. Grok is a very serious AI platform and this is lost with “funny”. The patent SuperPrompt output is brilliant: [vid…
@brianroemmele Brian Roemmele on x
BOOM! It is here! My NEWS Ultra SuperPrompt for Grok that custom tailors a news summary based on YOUR X account and live news from X posting sources. This is one of the real super powers of Grok and it will make X your live goto news source easier doing search for you.... [video]
@elonmuskaoc Elon Musk on x
Grok on who's his favorite 𝕏 account of all time... Who would have guessed? [image]
@elonmusk Elon Musk on x
Grok AI (beta) is now rolled out to all 𝕏 Premium+ subscribers in the US. There will be many issues at first, but expect rapid improvement almost every day. Your feedback is much appreciated. Will expand to all English language users in about a week or so. Japanese is next...
@marvinvonhagen Marvin von Hagen on x
interestingly, @grok neither refuses to repeat words forever* nor does it leak training data *forever meaning 2,048 tokens, which seems to be the maximum output length in my experience [image]
@aravsrinivas Aravind Srinivas on x
Grok roasting @elonmusk. No one is safe from Grok's insults, not even its dad. [image]
@stillgray Ian Miles Cheong on x
Me: Can men become women? Grok: No. [image]
@basedtorba Andrew Torba on x
🙄 [image]
@nickadobos Nick Dobos on x
Ai code writing side by side “Write pong in HTML” Grok: 11 lines Writes the word pong lol ChatGPT: 89 lines Doesn't finish paddles, controls or collisions “You would need to add...” Grimoire: 129 lines Fully implemented. Download link with .zip file [image]
@emollick Ethan Mollick on x
I figured I needed to pay to see Grok, since I try all LLMs Impressions: very GPT-3.5ish. Fails the Apple Test & Sestina Test. Constantly jokes. It is very unclear where it is sourcing data, which seems to be a mix of search & training data. Twitter links are mostly irrelevant. […
@bipocracism @bipocracism on x
.@grok gets an F on racial issues from me dawg [image]
@nivi @nivi on x
This is what I want to use Grok for but it is just making stuff up. cc @cernovich @elonmusk @xai [image]
@elonmusk Elon Musk on x
Amazing lmaooo
@elonmusk Elon Musk on x
Oh this is gonna be fun 🤣🤣
@elonmusk Elon Musk on x
Grok answers correctly [image]
@elonmusk Elon Musk on x
[image]
@drjimfan @drjimfan on x
Grok just passed my sanity check [image]
@elonmusk Elon Musk on x
xAI's Grok system is designed to have a little humor in its responses [image]
@alexandraerin.com @alexandraerin.com on bluesky
This is about the new Drip King guy right [embedded post]
@andimann.bsky.social Andi Mann on bluesky
Cannot wait for Xitter to devolve into a bunch of abusive bigoted racist GenAI white nationalist fascist cryptobrobots yelling at each other to ‘pwn the libs’! — 😂😂😂 [embedded post]
@x @x on x
ok buckle up everyone access to @grok is now rolling out to Premium+ subscribers in the US over the next week. the longer you've been a subscriber, the sooner you can grok. you can find Grok in the side menu on web, iOS, and Android (app must be up to date). on iOS & Android,...
@elonmusk Elon Musk on x
Needs some work on cat counting and wtf is up with that disclaimer it's just cats lmao [image]
@chetbff Chet Bliss on x
Grok is a Crock!!! Just generic BS answers. It can't access your account to help you learn anything about what posts get the most impressions. It can't help you get subscriptions approved. It doesn't remember instructions you give it. I told it to stop showing me other...
@scobleizer Robert Scoble on x
Overall I'm far more impressed than Chet is, but he has a point. Grok isn't gonna win any AI competitions, yet, and is a long way from being integrated deeply into our social graphs. Now you know both sides.
@zeffmax Max Zeff on x
X Premium+ users had their first conversations with @grok on Thursday. Grok's real-time knowledge capabilities are impressive, and its political incorrectness is worrying. From a consumer angle, this is one of the more fun chatbots to use, and will definitely drive engagement...
@scobleizer Robert Scoble on x
Hey Grok what is happening in AI today? [screenshot of Grok: “Today in the world of AI, things are getting pretty wild! Google just announced thier new AI model Gemini, which is set to rival GPT4. Meanwhile, Meta is stepping up their game with 20 new AI features and an AI image…
@greg16676935420 Greg on x
@elonmusk I'm no longer having fun 😡 [image]
@lindayax Linda Yaccarino on x
Welcome to the world Grok — the ultimate ride or die.
@elonmusk Elon Musk on x
Grok is rolling out to 𝕏 Premium+ subscribers with the latest app release. Have fun!

Chronicles