Over the weekend, Elon Musk's startup xAI released its first AI model, called Grok, to a select group of users — and judging by screenshots posted by X-formerly-Twitter employees, the model will be a far more crass and dad joke-prone chatbot compared to more professional offerings like ChatGPT.

After X employee Christopher Stanley asked the bot when it's OK to listen to Christmas music, for instance, Grok promptly told him that anybody who complains should "shove a candy cane up their ass and mind their own damn business."

In other words, it's quintessentially Elon Musk, who's always had a strong penchant for trolling behavior and calling his critics names.

Last week, Musk promised that the model would be released to a "select group" as soon as this past weekend, claiming that "in some important respects, it is the best that currently exists."

It's exactly the kind of vague and overpromising statements you'd expect the mercurial CEO to make. And given what we've seen, those "important respects" amount to allowing an AI model to exist in a "fucking free-for-all," swear words and all.

Who asked for this?

Musk's AI is essentially a response to what he sees as excessively "woke" AI chatbots that have been playing it too safe. Musk went on a tirade earlier this year, arguing that companies like OpenAI are actively censoring their models and giving in to the purported demands of the left.

Of course, one motivation behind the latest AI could be that he rage quit OpenAI, which he co-founded, over fundamental disagreements with the company's leadership back in 2018 and is now trying to secure a slice of the AI pie.

So is Grok more in line with what Musk had in mind? A brief perusal of the company's documentation, which could've easily been written by Musk himself, suggests it's just that.

The AI is "designed to answer questions with a bit of wit and has a rebellious streak, so please don’t use it if you hate humor!" the company's website reads.

Worst of all, the AI "has real-time knowledge of the world via the 𝕏 platform," suggesting it'll lean on extremely skewed and fundamentally untrustworthy data. "It will also answer spicy questions that are rejected by most other AI systems."

X's vague documentation and the latest screenshots leave plenty of questions unanswered. For one, we still don't know if this kind of aggression will be part of the regular chatbot or if there'll be an opt-in "fun mode," as one recently shared clip suggests.

Then there's the question of why Grok exists in the first place. Is it meant to serve as a way to entertain by coming up with racially tinged and dated humor and "sarcasm" — or is it actually there to inform X users of what is happening in the world?

Considering the current state of the discourse on Musk's social media platform — the dataset being fed to Grok — there's clearly little to learn about reality. To be fair, xAI's many much larger competitors have also been struggling with getting their AI models to stick to the truth.

Musk has publicly dreamed of building a "maximum truth-seeking AI" called "TruthGPT" that doesn't bat an eyelid if its users ask it how to make cocaine or build explosives. But judging by the sheer amount of misinformation being spread on X, an issue metastasizing in a near vacuum of moderation or guardrails, Grok will be anything but "truth-seeking."

In short, unless xAI somehow manages to transcend the capabilities of all of its far more established rivals and turn tweets into a reliable source of information, there's a good chance Grok will turn out to be a tired dad joke generator aimed at those who are willing to continue spending time on Musk's platform.

But even that isn't exactly a desirable outcome, given the growing issue surrounding the spread of misinformation.

Laughing at Musk's low-effort jokes is one thing — but equating those jokes with the truth is something entirely different.

More on Musk's AI: Elon Musk Is About to Release His First AI


Share This Article