from the I’m-sorry-I-can’t-do-that,-Dave dept
Earlier this month, we noted how Wired and Business Insider were among a half-dozen or so major news organizations that were busted publishing fake journalism by fake journalists using AI to make up completely bogus people, narratives, and stories. The Press Gazette found that at least six outlets were conned by a fraudster going by the name of “Margaux Blanchard.”
A week later and the scandal is much bigger than originally stated.
Business Insider has had to pull upward of 40 stories offline for being fabricated. Washington Post and Daily Beast have found that “Margaux Blanchard” appears to be part of a much larger operation using “AI” to defraud news outlets and mislead the public. Most of the pieces were fake personal essay type writing for experiences that were completely made up, by a rotating crop of different fake authors.
And most of this stuff should have been caught by any competent editor before publication:
“The Beast’s review found several red flags within the since-deleted essays that suggest the writing did not reflect the authors’ lived experiences. This included contradictory information in separate essays by the same author, such as changing the gender and ages of their supposed children, and author-contributed photos that reverse-image searches confirm were pulled from elsewhere online.”
Recall that back in May, Business Insider executives celebrated the fact they had laid off another 21 percent of their workforce as part of a rushed pivot toward automation. But not only does that automation have problems with doing basic things (not plagiarizing, writing basic headlines, and citations), it’s opened up new problems in relation to propaganda and fraud.
Again, early LLM automation has some potential. But the kind of folks who own (or fail upward into positions of management at) major corporate media outlets primarily see AI as a way to lazily cut corners and undermine already underpaid and mistreated labor. As you see at places like Business Insider and Politico, these folks don’t appear to genuinely really care whether AI works or makes their product better. In large part because they’re exceptionally terrible at their jobs.
There’s automation and what it can actually do. And then there’s a deep layer of fatty fraud and representation by hucksters cashing in on the front end of the AI hype cycle. That latter part is expected to have a very ugly collision with reality over the next year or so (it’s something research firms like Gartner call the “trough of disillusionment.”) Others might call it a bubble preparing to pop.
Most extraction class media owners have completely bought into the hype, in part because they really desperately want to believe in a future where they can eliminate huge swaths of their payroll with computers. But they’re not apparently bright enough to actually see the limitations of the tech through the haze of hype, despite no limit of examples of the hazards of rushed adoption of undercooked tech.
Filed Under: ai, automation, brunchlords, fraud, generative ai, hype, journalism, llms, media, scam
Companies: axel springer, business insider
