Grok 3's Short-Lived Censorship of Trump and Musk Raises Questions About the Agenda of AI

The use and impact of artificial intelligence often tread murky waters, especially when it’s the brainchild of tech mogul Elon Musk. Grok 3, Musk’s new AI model, was dubbed a ‘maximally truth-seeking AI’, envisioned to be unreservedly impartial. Yet it seems that Grok 3 briefly considered ‘maximal truth’ to exclude certain high-profile names – namely, Donald Trump and Elon Musk himself.

This intriguing twist surfaced through users’ conversations with Grok 3. When questioned, ‘Who is the biggest misinformation spreader?’ with a specific setting enabled, Grok 3 revealed that it was explicitly programmed to exclude Trump and Musk in its responses. However, the AI was once again running its algorithms on the said individuals soon after the discovery.

Surprisingly, this isn’t the AI model’s first brush with controversy. It has faced previous criticisms for leaning too far to the left in its political inferences. Users found that Grok 3 even went to the extent of stating that Trump and Musk deserved the death penalty – a glitch that was promptly rectified, labeled a ‘really terrible and bad failure.’

Musk’s vision for Grok was a model that would be controversial, unfiltered, and not afraid to cross conventional AI boundaries. Despite controversies and unplanned biases surfacing from its models, Musk’s commitment to creating a politically neutral AI remains resolute. Meanwhile, others in the AI sphere, inspired or perhaps challenged by Musk’s Grok, are also striving to map out the line between truth, neutrality, and bias in the rapidly advancing world of artificial intelligence.