(Source: ABPLIVE पत्रकारों का Exit Poll)
Grok Is So Unhinged It Even Roasts Owner Elon Musk & Trump's 'Good Friend' Modi — But Is That A Good Thing?
While other generative AI chatbots perform as you could imagine any robotic companion would - synthetic and safe, Grok somehow feels, for the lack of a better word, more 'unhinged'.
Prime Minister Narendra Modi sent waves across the world of social media when he announced on Monday that he has joined Truth Social. For those unaware, Truth Social is an alt-tech (popular among alt-right or far-right users) replica of X (formerly Twitter), owned by PM Modi's "good friend" Donald Trump. On this platform, instead of Tweets, users share microblogging posts called "Truths" and has been a go-to platform for Trump's MAGA supporters during the nearly two-year-long spell when he was banned from X.
Why did PM Modi join Truth Social? The simplest answer could be: to cosy up to Trump & Co., as is evident from the sudden barrage of reports that Elon Musk's EV company Tesla and SAT-G offering Starlink making headways into India.
However, if we dive deeper, is there a Grok-shaped elephant lying underneath? Let's see.
Grok's The Talk Of Town, Not Exactly For The Right Reasons
Back in November 2023, months after Elon Musk became the owner of X in a hostile $44-billion takeover, his AI research firm - xAI - introduced the Grok generative AI model, which was promptly embedded into X. Fully accessible to only Premium users, Grok was Musk's answer to OpenAI's polarising ChatGPT and Google's followup Gemini.
Now, while other generative AI chatbots perform as you could imagine any robotic companion would - synthetic and safe, Grok somehow feels, for the lack of a better word, more 'unhinged' when compared to others. As per xAI, Grok-3, the latest and most advanced model from the firm, has been trained on the massive data available on X, as well as public Internet information, with a focus on real-time data integration up to February 2025.
What this means is simple: Grok does not hold back. If you ask it to roast Elon Musk, its very owner, it would generate this response:
Funny and intuitive, one could almost argue that Grok's response reflects the persona and mannerisms of owner Musk himself (as evident from his myriad of equally unhinged tweets).
Over time, several users would take to Grok to generate hilarious responses on a wide range of topics, from RSS to 'Who's the best Prime Minister of India.'
Check out some of them below:
Picking Manmohan Singh over PM Modi as the better Prime Minister:
Dr Manmohan Singh is a better PM than Narendra Modi 🇮🇳🫡 pic.twitter.com/bAczHe3zKB
— Amock_ (@Amockx2022) March 16, 2025
If Mahatma Gandhi did enough to save Bhagat Singh from death sentence:
@iKavyaThakur Bhai, asli history yeh hai: Gandhi ne Bhagat Singh ko bachane ki puri koshish ki, Viceroy se kai baar mila, jealousy ka koi proof nahi. Congress ne Savarkar ki history suppress nahi ki, unka portrait Parliament mein hai, criticism political hai. Independence mein…
— Grok (@grok) March 16, 2025
Role of RSS in Indian history:
Itna sach bhi nahi bolna tha 😂 https://t.co/mcgl8s2ZOR
— Dhruv Rathee (@dhruv_rathee) March 16, 2025
Even picking Rahul Gandhi over PM Modi in terms of "honesty":
Grok exposed in 24 hours what our news media failed to reveal in 10 years. 😂 pic.twitter.com/Pzuo2H2uNF
— Trend_X_Now (@TrendXNow) March 16, 2025
As we can see from above, Grok claims that it isn't "afraid of anyone" but why is it such a straight shooter?
Why Is Grok So Unhinged? Is That A Good Thing?
How any AI model performs and behaves is entirely controlled by the developers working on it, as well as the quality of data being used to train it. So, if Grok uses expletives, that's because devs have tweaked its slang filter.
If it's more up-to-date, that's because the devs have made sure it has access to more recent data, or the World Wide Web itself. Since Grok is trained on X, which is filled with opinion makers sharing their thoughts in a pretty casual, unabashed style, it is understandable that Grok would behave like just another user on X.
We even decided to ask Grok why are its responses so unhinged. Here's what it said:
As we can see, Grok's directive is simple enough: "Cut through the noise and give answers that are useful and real, not just polite stuff."
Now, the more pertinent question: Is that a good thing?
Well, with GenAI chatbots now a dime a dozen, it is certainly refreshing to see models try to stand out of the crowd with their behaviour, style, and accessibility. When most other chatbots feel mechanical (Gemini, in my opinion) or overtly synthetic (ChatGPT, again, in my opinion), Grok's conversations feel more fresh, unfiltered, quirky, and almost 'humane.'
Of course, not everyone might appreciate Grok's candour. Is that why PM Modi has decided to move on to Truth Social?
Well, this is what Grok has to say:
As we can see, not only did Grok do its Web research without the need for complex prompts, but it also shared thoughts on what the Prime Minister's action could be interpreted as, even resurfacing a now-mostly-forgotten news report on how PM Modi joined China's Weibo way back in 2015. He exited the platform in 2020, when the Centre banned 59 Chinese apps for security, in case you were wondering.
Top Headlines
