“We plagiarized your plagiarism so we could put plagiarism in your plagiarism.”
Soft Launch
Elon Musk’s Grok AI was already having a rough launch, with the bot trashing Musk and cosigning a bunch of progressive political causes that are anathema to the increasingly regressive entrepreneur.
Now, add another woe to Grok’s rocky debut: users are noticing that it seems to be cribbing from its direct competitor ChatGPT, which is made by Musk’s former pals and current enemies at OpenAI.
In response to one query, for instance, Grok made a startling admission: “I’m afraid I cannot fulfill that request, as it goes against OpenAI’s use case policy.”
Remember, OpenAI didn’t make Grok — Musk’s xAI startup did, at least in theory. So what’s going on?
Excuse Goose
An xAI engineer named Igor Babuschkin quickly weighed in to offer an explanation.
“The issue here is that the web is full of ChatGPT outputs, so we accidentally picked up some of them when we trained Grok on a large amount of web data,” he wrote. “This was a huge surprise to us when we first noticed it.”
Whether or not that’s true, it’s increasingly well established that weird stuff does start to happen when AI is trained on the outputs of other AI. And it’s not the world’s least plausible excuse, either, because we’ve already seen Google’s AI vacuuming up and regurgitating the work of ChatGPT.
“For what it’s worth, the issue is very rare and now that we’re aware of it we’ll make sure that future versions of Grok don’t have this problem,” Babuschkin continued. “Don’t worry, no OpenAI code was used to make Grok.”
If that sounds like there wasn’t much testing on Grok before releasing it to the world… well, yes. As such, observers were quick to drag the excuse.
“We plagiarized your plagiarism so we could put plagiarism in your plagiarism,” quipped NBC News reporter Ben Collins.
More on Grok: Elon Musk Seeking $1 Billion for His Potty-Mouthed AI
