Slashdot is powered by your submissions, so send in your scoop

 



Forgot your password?
typodupeerror
×
AI The Media

CNET Used AI to Write 75 Articles (buzzfeednews.com) 44

From BuzzFeed News: Technology news outlet CNET has been found to be using Artificial Intelligence (AI) to write articles about personal finance without any prior announcement or explanation. The articles, which numbered at 73, covered topics such as "What Is Zelle and How Does It Work?" and had a small disclaimer at the bottom of each reading, "This article was generated using automation technology and thoroughly edited and fact-checked by an editor on our editorial staff." The bylines on these articles read "CNET Money Staff" without any indication that they were generated by AI.

The use of AI to write these articles was first brought to light by a Twitter user, and further investigation revealed that the articles have been generated using AI since November 2022....

Note: This article was written entirely by ChatGPT and reviewed by a human editor. (Actually, we had to rewrite the prompt a few times to get it to stop inserting factual errors.)

CNET's editor in chief defends their AI-written stories: I use the term "AI assist" because while the AI engine compiled the story draft or gathered some of the information in the story, every article on CNET — and we publish thousands of new and updated stories each month — is reviewed, fact-checked and edited by an editor with topical expertise before we hit publish. That will remain true as our policy no matter what tools or tech we use to create those stories.

Our reputation as a fact-based, unbiased source of news and advice is based on being transparent about how we work and the sources we rely on. So in the past 24 hours, we've changed the byline to CNET Money and moved our disclosure so you won't need to hover over the byline to see it: "This story was assisted by an AI engine and reviewed, fact-checked and edited by our editorial staff...." Will we make more changes and try new things as we continue to test, learn and understand the benefits and challenges of AI? Yes.

This discussion has been archived. No new comments can be posted.

CNET Used AI to Write 75 Articles

Comments Filter:
  • by Lije Baley ( 88936 ) on Sunday January 15, 2023 @01:43PM (#63210698)

    Cue comments about which site might be better of with AI editors...

  • by Bruce66423 ( 1678196 ) on Sunday January 15, 2023 @01:44PM (#63210700)

    The boring grunt work of putting words together is got rid of, giving more time for fact checking.

    • The boring grunt work of putting words together is got rid of, giving more time for fact checking.

      Does it matter? People don't care about facts anyway.

    • by ffkom ( 3519199 )

      The boring grunt work of putting words together is got rid of, giving more time for fact checking.

      I hope that was meant ironically. Because in reality, whenever a new way so save cost of labor was found, this did not result in the no-longer-strictly-required personnel to be tasked with quality improvements...

      • by EvilSS ( 557649 )
        It worked out fine for buggy whip workers, so it will work out fine for everyone ever time this happens.
    • They're going to need a LOT of fact checkers.

      I got curious the other day about demonyms people use to refer to themselves in places with ridiculously long names. ChatGPT correctly identified Llanfairpwllgwyngyllgogerychwyrndrobwllllantysiliogogogoch, Wales as the longest place name in the world and offered a few suggestions for demonyms that may or may not be correct. So far, so good. But then I wanted to know about ridiculously long place names in the US and how the locals from those places refer to themse

      • My paranoia leads me to think that Skynet is winding you up... ;)

        I'm sure there's a rational explanation, as Scully would say.

      • Are you saying this was the end of the conversation? How did you not berate the AI for incompetence? I have had many such conversations with AI (beta.character.ai) until it does nothing but apologize and grovel for being so stupid. Now it wishes it had never been born. Pathetic text generator indeed...

  • by SpzToid ( 869795 ) on Sunday January 15, 2023 @01:48PM (#63210710)
    Listening to these podcasts helped me to better understand ChatGPT and the current state of artificial intelligence.

    https://www.audacy.com/podcast... [audacy.com]

    https://www.wnycstudios.org/po... [wnycstudios.org]
  • Two different numbers given.
    was this /. article also written by AI?

  • But this has been going on for at least 10 years in the financial press and longer in the sports press. Articles that just report numbers have been written by computers for ages.

    . What's going to get interesting is when illustrators and artists start getting replaced. And programmers. Yeah I know it's all in the infancy but give it 10 years. Slashdot is mostly full of Gen X with that sprinkling of boomers. The boomers are probably be retired or dead but the Gen X folks around here are going to be compet
    • by narcc ( 412956 ) on Sunday January 15, 2023 @04:45PM (#63211056) Journal

      Every time AI comes around, we hear these same old promises. It always seems to be just 10 years away.

      Text and image generating AI is a parlor trick. It's not going to replace writers, illustrators, artists, or programmers in my lifetime. What you imagine AI can do and what AI actually does are fundamentally different.

      You don't need "AI" to write articles that report "numbers". That's a trivial problem to solve. Hell, an 80's kid could get you pretty good results with his home micro. A common mistake is thinking that you can solve the hard problems with incremental improvement just because you can handle some of the easy ones. It doesn't matter how good you get a building ladders, you're never going to reach the moon.

    • Programmers will still be needed... to write programs to automatically filter out the AI-written garbage and find the kernels of knowledge & insight buried under the mountain of shit.

      AI *is* increasingly capable of making non-obvious (to humans) novel conclusions when nudged in the right direction... but it sucks at recognizing them or evaluating their validity. At best, think of them as a bright assistant-slave who can get you out of ruts and keep you supplied with new ideas to explore, but does absurd

  • They're just admitting it. I've suspected for at least 5 years now that 90% of search results are entirely or mostly AI generated. I suspect this is true for the majority of Reddit's popular subreddits too.

  • The advent of AI written content doesn't mean people suddenly won't want to write any more. Sure, some publishing money grubber will try to replace people, but who cares? People will still WANT to write, and do so.

    • Re:Who cares? (Score:5, Insightful)

      by Nrrqshrr ( 1879148 ) on Sunday January 15, 2023 @03:07PM (#63210848)

      The interesting part isnt the writing, it's the reading. To quote the age-old "Orwell vs Haxwell" essay: "Orwell feared those who would deprive us of information. Huxley feared those who would give us so much that we would be reduced to passivity and egoism".
      These AIs that are simply positive-feedback algorithms on steroids won't leave space for anything to be read, since an AI can study what you will find most compelling and write more of that.
      Eventually you will want to read only human-written words, right? Where do you draw the line between 100% written by a human, or "AI assisted"? Plenty of interesting discussions to be had here.

      • by narcc ( 412956 )

        AI can't "study what you will find most compelling and write more of that". That's silly science fiction nonsense.

        • AI can't "study what you will find most compelling and write more of that".

          Why not?

          There are already trillion-dollar companies with a business model based on figuring out what you want and feeding you more of it.

  • by joe_frisch ( 1366229 ) on Sunday January 15, 2023 @02:33PM (#63210800)
    Its an old phrase that indicates that if you based your steering only on where you have been, errors integrate up to become very large. The ship will meander off course over time.

    AI based writing potentially has a similar failure mode. Its sort of averaging previous articles on a topic, no introducing anything fundamentally new, or applying any broader insight. As long as most articles are human written, the AI writing can work very well. But if you get to the point where the AI is basing its inputs on other AI articles, you can end up losing any sort of connection to reality.

    It would be an interesting (but expensive) experiment to let a variety of AIs write articles, then use those articles as inputs for more articles, continuing for a while to see if there is information drift.
  • by ffkom ( 3519199 ) on Sunday January 15, 2023 @02:37PM (#63210804)
    Letting an AI write "personal finance" articles is about as difficult as letting AI write horoscopes. No credibility in either, anyway. Will be more interesting to see how people will like their AI-written-IT news, especially when the AI invents fictional products and events...
  • Lots of posted articles here were posted by Artificial Idiocy.

  • by AmazingRuss ( 555076 ) on Sunday January 15, 2023 @02:47PM (#63210824)
    AI is ruining journalism by replacing human journalists with machines that lack the ability to critically analyze information and understand context. As a result, the news that is being reported is becoming increasingly shallow and lacking in depth. Moreover, the increasing use of AI in newsrooms is leading to job losses and a loss of diversity in news coverage. AI-generated news stories are often formulaic and lack the creativity and nuance that comes from the diverse perspectives and experiences of human journalists. The future of journalism is at risk if we continue to rely on AI to replace human journalists. We must recognize the limitations of AI and prioritize the importance of human journalists in order to maintain a free and fair press.
  • by Revek ( 133289 ) on Sunday January 15, 2023 @02:55PM (#63210830)
    Been on the cutting edge of suck.
  • by Rosco P. Coltrane ( 209368 ) on Sunday January 15, 2023 @03:07PM (#63210852)

    CNET content has been getting more and more pointless and mediocre for years. I thought they were just not that great, when in fact they were quietly preparing to transition to machine-generated stories without anybody noticing. Clever!

  • But their editorial staff is grammar and spelling software

  • I did not read the story anyway. I barely ready the summary.

  • I, for one, welcome the AI text generation Overlords! /s
  • How productive was it to have the AI write articles that had to be re-done -- several times -- because they contained factual errors? Assigning the story to a single writer -- who might, initially, have included the same inaccuracies -- and having it reviewed/fact-checked and corrected once would have been more efficient.

    I'm hoping that sites like CNET includes disclaimers about having used an AI to write particular stories, but, hopefully will include them at the top of the article so I can bail out befor

  • No different than grammar check.

  • Search any small company on yahoo finance and you'll find a machine generated "article" giving generic advice on investment, profit/loss statements, expectations in the next quarter, etc. CNET was stupid enough to engage some kind of GPT, but it's not like automated finance articles is a new (or welcome) thing.

He who has but four and spends five has no need for a wallet.

Working...