Google’s Bard AI chatbot has been caught plagiarizing a CPU review from Tom’s Hardware and it doesn’t seem happy about it. A journalist at the tech news website asked Bard, “which of two competing processors — the Intel Core i9-13900K or AMD Ryzen 9 7950X3D — was faster,” and that’s when things took an interesting turn. It appears that the AI essentially scrapes the internet for data in order to answer questions. The problem is that it doesn’t necessarily cite its sources when doing so, that is unless the user presses the AI to provide them.
From Tom’s Hardware:
“The answer it gave was taken directly from one of our Tom’s Hardware articles, but Bard didn’t mention the article and instead referred to the number as occurring “in our testing,” implying that Google itself had done the benchmarking.”
Upon further questioning about where the AI obtained its information from it then stated Tom’s and when asked if it had just plagiarized by not citing its source, it said yes. Bard even went so far as to apologize for its mistake and that it would try to be more careful in the future. However, when the journalist provided a screenshot of the previous conversation and asked Bard, a day later, if it had ever committed plagiarism the AI responded that the screenshot was a fake. Paul Lilly (Hot Hardware) also confirmed getting a similar response.
Beta Does Afterall Mean Not Quite Ready for Primetime
Google’s Bard AI chatbot is technically in a Beta phase and Google itself has said that it is an “experimental conversational AI service”. Google has even gone so far as to include disclaimers on Bard’s homepage clarifying that the service “may give inaccurate or inappropriate responses.” The Bard FAQ does include a section regarding how and when Bard cites its sources.
From Bard FAQ:
“Bard, like some other standalone LLM experiences, is intended to generate original content and not replicate existing content at length. We’ve designed our systems to limit the chances of this occurring, and we will continue to improve how these systems function. If Bard does directly quote at length from a webpage, it cites that page.
Bard was built to be a creative and helpful collaborator—it works well in creative tasks like helping you write an email or brainstorm ideas for a birthday party. We see it as a complementary experience to Google Search. That’s why we added the “Google It” button to Bard, so people can easily move from Bard to explore information from across the web.
Bard is an experiment, and we’ll use its launch as an opportunity to learn, iterate, and improve the experience as we get feedback from a range of stakeholders including people like you, publishers, creators, and more.“
What happened regarding the instance of it using data from Tom’s CPU review does somewhat reflect what is stated here but clearly, it needs a bit more work. Tom’s further clarified about the responses Bard gave vs. what was in their review which shows how the AI honed in on key data points for its answer.
“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.”
– Tom’s Hardware
“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.”
Bard
Discussion (8 replies)
Join Discussion →The review itself was probably written by a chatbot.
I wonder...... I've played with bard some... it's a fun tool.
I wonder . . .years ago I knew an IT person whole told a story about someone getting Alexa (on an android phone) and Siri (iPhone of course) in some kind of conversation with each other, what might happen if we can get ChatGPT and Bard to interact?
Isn't plagiarizing essentially ALL that AI does?
It searches th einternet and then semi-intelligently sews what it finds together. The end result is plagiarizing its internet search.
I mean isn't what what most modern non original research papers are?
Yup, but as long as you include a citation it isn't plagiarizing - then it's just "research"
There are lines.
In college, and it wasn't for journalism but did include classes on writing (I apologize to all who've tried to guide me but my various stupors overpowered), there were some pretty clear lines about citing sources. Meanwhile, on the net, there is a bit more of a grey line and I've tried my best to find a middle ground. Personally, I feel that, at the very least, provide a link to what you're sourcing. As someone who has worked hands-on with tech and then had to troubleshoot it, while bridging the gap between it back to human users who have minimal tech skills, I've been able to see where programming deficiencies have happened. I admit, I use AI all the time to help restore (not necessarily correct), what I should be expressing. I say that because what it often suggests are things I know, but at post-middle age, that I just want to ignore. In many cases I accept it but others I'm, like, f* it this is what I want to say. If an AI is going to scrape info, it should give some kind of credence to the source. Btw this entire post was written by me.
For sure, but Google makes money by rankings of searches, the ad models support that, and if it can find an end-run to escape that then everything on the net it uses as a source is exempt from that model.