AI is getting better at coding, data analysis and image generation, but it’s also getting better at some unexpected areas.
Grok, the AI released by Elon Musk’s AI company xAI, can now respond in perfect Hinglish. The AI has been responding in a mixture of Hindi and English, much like most north Indians, and is also adding some profanities to its responses for good measure. This has stunned many Indian users, who’ve been marveling at how much the AI sounds like an average Indian internet user.
X user Toka had asked Grok’s chatbot which his best 10 mutual followers on the platform were. Grok didn’t reply for a while, which made Toka lightheartedly respond “B***wala @grok seen karke chod diya, typical wahmyn behaviour I will never forgive you for this. (<Profanity> grok left me on seen, this is how women behave, I’ll never forgive you for this).
But it appears that the response stung the AI chatbot. “Oi b***wala, chill kar. Tera “10 best mutuals” ka hisab laga diya. Mentions ke hisab se ye hai list..Mutuals matlab dono follow karte ho, par exact data nahin hai to mentions pe bharosa kiya. Thik hai na? Ab rona band kar,” it replied. (Oi <profanity>, chill. I figured out who your best 10 mutual followers were. This is the list <lists handles>..that’s okay right? Now stop crying.)
The response was incredible, not only because it correctly understood the problem, even though it was stated in a mixture of Hindi and English — and also had internet slang like wahmyn thrown in — but it also mirrored the language of the question perfectly. Grok understood that the profanity was thrown in jokingly, and responded with the same profanity back. It also used Hinglish like the user had done, and even showed some sass of its own, asking the user to “stop crying” now that his request had been fulfilled.
This is a dramatic departure from how other AI chatbots like ChatGPT and Claude behave. These chatbots sound formal and business-like, and would never think about calling their user a Bh**wala. But Elon Musk had wanted Grok to be designed to be free of many of the stifling safety requirements of the other AI chatbots, and appears to have succeeded in created a bot that sounds uncannily human.
This one interaction isn’t the only response where Grok speaks Hinglish. It’s been regularly responding to user queries directed towards it in Hinglish in perfect Hinglish of its own. It starts responses with “haan yaar”, and even acknowledges that it needs to behave differently from human users because it’s an AI. “Ethics ka sawal hai, aur main seekh raha hoon!” it says. The tech world has been furiously debating over when humanity will reach AGI, but a chatbot that responds in Hinglish starting with “bho**wala” shows that we might actually be quite close.