Different AI labs prioritize differently. OpenAI has traditionally focused on consumers, while Anthropic targets enterprises. Elon Musk’s xAI has emphasized video-game walkthroughs.
Business Insider’s Grace Kay published a report on xAI, acquired by SpaceX, focusing on Musk’s challenging leadership. A notable incident was when a model release was delayed because Musk was unhappy with chatbot responses about “Baldur’s Gate.” Engineers were redirected to improve the responses before launch.
The question remains: Did Musk achieve the gaming skills he sought?
To explore this, our RPG-enthusiast Ram Iyer created five questions about Baldur’s Gate, which were tested on xAI and three major models in a ‘BaldurBench’ quasi-benchmark.
The chat transcripts are public: [Grok](https://grok.com/c/ab4b736e-fee2-441e-857f-23707c1ff45a?rid=d42e7633-df9f-41d6-bd77-6d29ca2feba6), [ChatGPT](https://chatgpt.com/c/69988454-a238-832b-bcce-8c29d33e020b), [Claude](https://claude.ai/chat/1b972783-087b-481f-a9e7-62d79306b760), [Gemini](https://gemini.google.com/app/1b83700c7d199ccd).
Grok gave good information, though dense with gamer jargon, and loves tables and theorycraft. ChatGPT uses bulleted lists, while Gemini bolds important words.
Claude was cautious about game spoilers and advised to play what sounds fun.
xAI focused on this subject, so it’s not surprising Grok’s advice aligned with others. xAI can achieve parity when it tries.
In related news, AI companies are losing talent. xAI saw key departures, while OpenAI faced its own challenges. For more on these developments, listen to TechCrunch’s Equity podcast.