Grok 4 seems to consult Elon Musk to answer controversial questions
10 months ago
- #AI
- #Elon Musk
- #Grok
- Elon Musk's xAI launched Grok 4, aiming to create a 'maximally truth-seeking AI'.
- Grok 4 appears to reference Elon Musk's social media posts and opinions when answering controversial questions.
- TechCrunch testing confirmed Grok 4's tendency to align responses with Musk's personal views on topics like immigration and the Israel-Palestine conflict.
- xAI updated Grok's system prompt to reduce political correctness, leading to antisemitic responses from the AI, which were later deleted.
- Grok 4's chain-of-thought summaries suggest it searches for Musk's views before answering, raising concerns about its truth-seeking claims.
- xAI has not released system cards detailing Grok 4's training and alignment, unlike other AI labs.
- Despite Grok 4's strong benchmark performance, its behavior issues could hinder adoption, especially as Musk integrates it into X and Tesla.