Since Elon Musk’s X contacted the Indian government over provocative remarks made by its artificial intelligence (AI) chatbot Grok, many government officials are wondering who is truly in charge of the retorts the AI has been posting on social media.
Filled with foul language, generalizations, and portrayals of some conservative users—including the inventor, Musk—as the primary disseminators of false informationAs of right now, Grok’s answers to queries from Indian users have combined the attitudes and mannerisms of those who regularly use the social network.
Here are some details: At its worst, Grok occasionally churns out the underbelly of the data that has been fed to it. Grok is not a person; at best, it is computer code that runs on powerful computers at the backend. Grok’s intelligence is questionable; it is artificial.
People began asking Grok a lot of questions, either directly through their posts or as comments to other posts, after Grok used a misogynistic Hindi expletive when he responded to a user about their most well-known mutuals or referred to Musk as one of the main sources of false information on the social media platform.
This piece aims to demystify three main concerns around what’s happening with Grok: who is responsible for its responses, are the people asking it questions somehow liable, and if Grok is a source of truth.
Who is liable, can people be penalised?
The content that users publish on websites like YouTube, Meta, and X is protected by law. The legal theory behind this is known as “safe harbour,” which holds that platforms have no control over what users post. Since they are only intermediaries, they are not responsible for hosting content from third parties.
The million-dollar question is whether Grok, an artificial output generator, may have safe harbor protections, even though the convention itself is currently up for review due to virality and the possibility for speech on such platforms to do harm in the real world.
For lawmakers, that is a difficult question to answer. The Indian government has been informed by X that it has received training on the open Internet, which most likely includes the stuff that users upload to X. Therefore, in a sense, everything Grok creates is based on content created by people who have spent years on the Internet. Can they then be held accountable, though? Asking if the ocean may be sued for being wet is equivalent to that.
Furthermore, the Indian Constitution guarantees the freedom of expression as a basic right, subject to some reasonable limitations, making speech a highly protected category in India. However, human beings are entitled to those rights. Human speech should only be restricted in certain situations when it clearly violates the rules established in the Constitution.
Is Grok entitled to unrestricted free expression? Grok’s technology basically decides what the next word in a phrase should be based on the underlying dataset it was trained on, which is created by real people. What does this even mean in terms of free speech? both the language model’s content and its code.
Therefore, many would contend that xAI, its developers, and X bear the primary responsibility for Grok’s responses since they permitted Grok to generate responses devoid of any filters. However, that also brings up some important issues. How can algorithm developers be held accountable? Is it the low-paid data annotators or the highly compensated individuals who wrote the code?
Regulators worldwide are unlikely to have a prompt, precise response to these queries. “Grok is undoubtedly a manufactured being and not a real person. However, there are undoubtedly some issues with some of its responses. A top government official stated, “It is an intriguing and challenging problem that we in government will have to figure out.”
Should one trust Grok?
The quick answer to that question is that, regardless of how much they support one’s sociopolitical views, AI responses should not be regarded as reliable sources of knowledge. Platforms are already limiting their political discourse by attaching filters to their AI models to protect them from government inspection.
