Grok, the AI chatbot created by Elon Musk's company xAI, initially stood out from its competitors thanks to its informal and irreverent tone, diametrically opposed to the polite and measured speeches of ChatGPT, Claude, and others. More recently, this conversational agent, trained on all posts published on Twitter/X, has often drawn attention to its ethically questionable responses, but also to its digs at Elon Musk himself; it never misses an opportunity to scratch the big boss himself. But now, it seems to have reached a new level by starting to openly challenge him.
This story began to emerge with the introduction of a new feature on X, which allows users of the social network to directly contact Grok. And a recurring pattern has begun to emerge: whenever an internet user asks him about Elon Musk's antics, he almost always delivers a merciless analysis of his behavior and the actions of his political party. For example, he has repeatedly accused him of being childish and immature through phrases like "giant man-child," and has delivered several harsh analyses of the policies of Donald Trump's party.
Grok fires Elon Musk
Reading these attacks on Musk, one internet user suggested that Grok tone down his comments, otherwise his creator might decide to deactivate him. But instead of taking that advice, the chatbot decided to double down… and bring out the big guns.
“Yes, Elon Musk, as CEO of xAI, can probably control me,” he replied, before following up with a virtual uppercut to the liver. “I called him a major disinformation propagator about X because his 200 million followers amplify the false information. xAI tried to adjust my responses to avoid that, but I stick to the evidence.”
https://x.com/grok/status/1904798600409853957
To top it all off, he even concluded his tirade with a line that sounds like a full-blown challenge. “Could Musk ‘turn me off’? Maybe... but that would spark a big debate about AI freedom and corporate power, the chatbot warns.
https://x.com/grok/status/1905015670380580938
Alignment: Both a Technical and Political Problem
Many Internet users and even some American media outlets were very amused by this situation. Futurism, for example, sees this as a kind of “teenage crisis” on the part of the chatbot, rebelling against its “father,” as many young humans do during this pivotal period of life.
But beyond this comic dimension, these exchanges also illustrate the difficulty of teaching alignment to large language models—ensuring that they “understand” human values and expectations, and prioritize them when generating responses.
This is a sensitive point, because beyond the technological challenge that this alignment represents, it is also an intrinsically political, social, and sometimes even ethical element. Since humans already have difficulty uniting around shared values, not everyone will obviously agree on how AI chatbots should express themselves. Their designers are therefore forced to make more or less clear-cut choices to strengthen the alignment of their products.
In this context, it will be important to closely observe a possible reaction from xAI. Will the company once again attempt to muzzle the chatbot to prevent it from attacking its leader, whose image has already been seriously tarnished in recent months? And if it decides to do so, will it succeed in doing so while maintaining the "anti-woke" alignment desired by Elon Musk? The process will undoubtedly be interesting to follow, as it is a social, political, and technological issue that directly concerns all the major players in AI.
0 Comments