Artificial intelligence laboratories often exhibit distinct strategic priorities. While OpenAI has historically concentrated on consumer-facing applications, its competitor Anthropic typically targets enterprise solutions. A recent revelation, however, indicates that Elon Musk’s xAI has placed a particular emphasis on aiding with video-game walkthroughs.
This insight emerged from a comprehensive report published last Friday by Business Insider’s Grace Kay. The report delved into xAI, the AI startup recently integrated into SpaceX, notably detailing challenges faced by employees under Musk’s leadership. Among the various accounts, one specific anecdote was particularly striking.
Sources familiar with the situation recounted an incident last year where a model release was delayed for several days. The reason: Musk expressed dissatisfaction with the chatbot’s ability to answer detailed questions about the video game “Baldur’s Gate.” Consequently, senior engineers were reportedly reallocated from other critical projects to enhance these specific responses before the model’s launch.
One can readily imagine the potential frustration for experienced engineers, accustomed to tackling fundamental challenges in knowledge and machine intelligence, finding themselves diverted to assist in a 54-year-old’s video game pursuits. However, this anecdote prompts an even more pertinent inquiry: Did Musk ultimately achieve the gaming proficiency he desired?
To investigate this question, our in-house role-playing game enthusiast, Ram Iyer, formulated a set of five general questions pertaining to “Baldur’s Gate.” We then subjected xAI’s model and three other prominent AI platforms to this assessment, in what we termed "BaldurBench," a quasi-benchmark designed for this specific purpose.
In the spirit of journalistic transparency, all chat transcripts from this evaluation have been made publicly available, featuring interactions with Grok, ChatGPT, Claude, and Gemini.
Commencing with the positive findings: Grok demonstrated a commendable ability to provide useful and well-informed information. While its responses were somewhat laden with gamer-specific terminology—such as “save-scumming” for saving and “DPS” for damage—the answers proved valuable for those familiar with the jargon. Grok also exhibited a notable preference for tables and detailed "theorycrafting," aligning with expectations for a gaming-focused AI.
Given the abundance of “Baldur’s Gate” guides available online, the models generally drew from similar source material. Consequently, the primary distinctions observed were stylistic. ChatGPT, for instance, favored bulleted lists and concise sentence fragments, whereas Gemini frequently utilized bold text to highlight key information.
The most unexpected outcome came from Claude, which displayed a distinct concern for avoiding spoilers that might diminish the user’s gaming experience. When prompted for advice on effective party compositions, Claude concluded its guidance with a reassuring remark: “don’t stress too much and just play what sounds fun to you.”
It is crucial to consider that, as reported by Business Insider, xAI has demonstrably focused on achieving proficiency in this particular subject area. Therefore, while Grok’s advice ultimately aligned closely with that of other models after the reported intensive development "sprint," it is important not to overinterpret this parity. Nevertheless, it affirms xAI’s capability to deliver effective results when its efforts are specifically directed.
The Editorial Staff at AIChief is a team of professional content writers with extensive experience in AI and marketing. Founded in 2025, AIChief has quickly grown into the largest free AI resource hub in the industry.