[ad_1]
\n<\/p>\n
This is your periodic reminder that AI-powered chatbots still make up things and lie with all the confidence of a GPS system telling you that the shortest way home is to drive through the lake.<\/p>\n
My reminder comes courtesy of Nieman Lab<\/em>, which <\/a> to see if ChatGPT would provide correct links to articles from news publications it pays millions of dollars to. It turns out that ChatGPT does not. Instead, it confidently makes up entire URLs, a phenomenon that the AI industry calls \u201challucinating,\u201d a term that seems more apt for a real person high on their own bullshit.<\/p>\n Nieman Lab<\/em>\u2019s Andrew Deck asked the service to provide links to high-profile, exclusive stories published by 10 publishers that OpenAI has struck deals worth millions of dollars with. These included the Associated Press<\/em>, The Wall Street Journal<\/em>, the Financial Times<\/em>, The Times<\/em> (UK), Le Monde<\/em>, El Pa\u00eds<\/em>, The Atlantic<\/em>, The Verge<\/em>, Vox<\/em>, and Politico<\/em>. In response, ChatGPT spat back made-up URLs that led to 404 error pages because they simply did not exist. In other words, the system was working exactly as designed: by predicting the most likely version of a story\u2019s URL instead of actually citing the correct one. Nieman Lab<\/em> did a similar experiment with a single publication \u2014 Business Insider<\/em> \u2014 earlier this month and <\/a>.<\/p>\n An OpenAI spokesperson told Nieman Lab<\/em> that the company was still building \u201can experience that blends conversational capabilities with their latest news content, ensuring proper attribution and linking to source material \u2014 an enhanced experience still in development and not yet available in ChatGPT.\u201d But they declined to explain the fake URLs.<\/p>\n We don\u2019t know when this new experience will be available or how reliable it will be. Despite this, news publishers continue to feed years of journalism into <\/a> in exchange for cold, hard cash because the journalism industry has <\/a> at figuring out how to make money without <\/a> to tech companies. Meanwhile, AI companies are <\/a> on content published by anyone who hasn\u2019t signed these Faustian bargains and using it to train their models anyway. Mustafa Suleiman, Microsoft\u2019s AI head, <\/a> anything published on the internet \u201cfreeware\u201d that is fair game for training AI models. Microsoft was valued at $3.36 trillion at the time I wrote this.<\/p>\n There\u2019s a lesson here: If ChatGPT is making up URLs, it\u2019s also making up facts. That\u2019s how generative AI works \u2014 at its core, the technology is a fancier version of autocomplete, simply guessing the next plausible word in a sequence. It doesn\u2019t \u201cunderstand\u201d what you say, even though it acts like it does. Recently, I tried getting our leading chatbots to help me solve the New York Times<\/em> Spelling Bee and watched them <\/a>. <\/p>\n If generative AI can\u2019t even solve the Spelling Bee, you shouldn’t use it to get your facts.<\/p>\n<\/div>\n