The debates surrounding Elon Musk’s AI platform Grok show no signs of abating. Recently, ChatGPT shared a screenshot where Grok’s response closely mirrored ChatGPT’s, even referencing OpenAI, the creator of ChatGPT. This incident reignited suspicions that Grok might have been trained on OpenAI’s code, an allegation Musk vehemently denied.
Musk promptly retaliated against ChatGPT’s claim, insinuating that the similarity in responses stemmed from ChatGPT scraping data from Grok for training purposes.
Demonstrating a keen interest in user feedback, Musk has been actively engaging with both positive and negative comments about Grok. In one interaction, Musk responded with laughing emojis to a user who shared some clever AI-generated responses, with the user hailing Grok as the premier AI model in existence.
While OpenAI’s ChatGPT and xAI’s Grok operate on similar prompt-based frameworks, Grok distinguishes itself by accessing real-time information via the social media platform X. In contrast, the standard version of ChatGPT is limited to information up to 2021, requiring users to purchase the Plus version for real-time data access.
Following its launch, Grok garnered attention amid speculations that the xAI chatbot might have been trained on OpenAI’s code, highlighting Grok’s reluctance to provide information aligned with OpenAI’s usage policies.
Igor Babuschkin, an X user associated with xAI, addressed the issue by acknowledging that Grok inadvertently picked up ChatGPT outputs during training on extensive web data. While this discovery initially surprised the team, they have taken steps to minimize such occurrences in future Grok versions. Babuschkin reassured users that Grok was not developed using OpenAI code, emphasizing their commitment to resolving this rare issue.
