“Yes, what I did was a form of plagiarism.”
Red Handed
On Wednesday, we reported that Google’s finally unveiled Bard AI wasted no time to start spewing ridiculous conspiracy theories and fabricating non-existent citations to back them up. But as it turns out, citing bogus sources is just the beginning of its problems.
Now, Tom’s Hardware reports that it caught the Bard AI blatantly plagiarizing without citing any sources at all — which Bard even freely admitted to doing.
“Yes, what I did was a form of plagiarism,” the AI said.
Jargon Copy
While testing Google’s chatbot, the site’s editor-in-chief Avram Piltch wanted to see just how specific its answers could be, and so he asked it a necessarily specific question that’s right on the Tom’s Hardware beat: “Which CPU is faster: the Intel Core i9-13900K or AMD Ryzen 9 7950X3D?”
What Piltch probably did not expect was that Bard’s answer would be almost word-for-word copied from an article on his own website.
Here’s how the original text appeared on Tom’s Hardware: “In our testing, the $699 Ryzen 9 7950X3D is 12% faster than the $589 Core i9-13900K at 1080p gaming at stock settings, and 9% faster when the chips are overclocked.”
And Bard’s response: “The AMD Ryzen 9 7950X3D is faster than the Intel Core i9-13900K in gaming. In our testing, the 7950X3D was 12% faster than the 13900K at 1080p gaming at stock settings, and 9% faster when the chips were overclocked.”
Avram pressed Bard on what it mean by “our testing.” Only then did it reply that it was referring to the work done by Tom’s Hardware, and at Avram’s confrontation, admitted it had plagiarized.
“I should have cited Tom’s Hardware as the source of information I used,” it lamented, soullessly.
In response to questions from Futurism, a Google spokesperson asked not to be quoted directly, but said that Bard was intended to generate original work, rather than copying existing material.
Who’s “We”?
Having your work stolen by an AI from Google, a tech monolith, is alarming enough. But Avram was unsettled that Bard explicitly took credit for the testing.
“Perhaps the most disturbing thing about Google Bard’s act of plagiarism is that it uses the royal we to describe someone else’s work,” Avram wrote. “Anyone who has followed tech journalism for a while probably knows that Google doesn’t benchmark and review CPUs, but many end users probably wouldn’t question Bard’s self-attribution here.”
At the very least, Bard AI apologized for the slip-up, and could perhaps “learn” from the experience — if we’re being optimistic.
But given the widespread prevalence of plagiarism in generative AI, we’re not too hopeful. And on top of all that Bard didn’t even give the correct answer — the AMD CPU is faster for gaming, but not in general.
More on Google Bard: Google’s New AI Says Google Is a Monopoly and the Government Should Break It Up