AI Fabricates More than Fake Legal Authorities

It can make up fake movie review quotes too.

Lionsgate pulled a trailer for the forthcoming Francis Ford Coppola movie, Megalopolis, this week. The trailer, which meant to show that some of Coppola's most well-regarded works were not always appreciated by movie critics at the time, contained multiple fabricated quotes.

Where did the fabricated quotes come from? Variety reports they were AI generated.

Sources tell Variety it was not Lionsgate or Egan's intention to fabricate quotes, but was an error in properly vetting and fact-checking the phrases provided by the consultant. The intention of the trailer was to demonstrate that Coppola's revered work, much like "Megalopolis," has been met with criticism. It appears that AI was used to generate the false quotes from the critics.

For instance, the trailer claimed that Pauline Kael wrote in the New Yorker that "The Godfather" was "diminished by its artsiness." Kael in fact loved the movie.

When Variety prompted AI service ChatGPT to provide negative criticism about Coppola's work from well-known reviewers, the responses provided were strikingly similar to the quotes included in the trailer.

It seems lawyers are not the only ones who need to be careful relying on the veracity of AI-generated sources.

Comments

[Frank Drackman]

Casablanca wasn’t reviewed all that well either (“Dark Passage” much better) or Lincoln’s Gettysburg Address (He said nothing about throwing Jefferson Davis in Jail! He kept looking at the back of an Envelope!)

Frank

[Kleppe]

The Gettysburg Address is a classic example of how fickle the American news media was (and remains). A shorthand reporter had been present on the platform at Gettysburg to transcribe Lincoln’s messaging in real time, and it was subsequently “telegraphed and printed in the leading newspapers of the country on the following morning.” The initial reaction in the domestic press was as muted as that of the audience following Lincoln’s remarks, which one witness likened to a “benediction”; similarly, a second witness recalled that “at the close the applause was not especially marked…” Philip M. Taylor wrote that the initial coverage of Lincoln’s messaging was “wholly inadequate”, observing one newspaper simply reported: “The President also spoke.” However, it wasn’t long before the American press changed tactics, either succumbing to the Band Wagon effect or social proof. In this case, several highly regarded English periodicals (including The Edinburg Review, The London Spectator, and The Saturday Review) spoke of Lincoln’s messaging in “the highest terms of commendation”, and the leading American newspapers soon followed suit. Notably, several domestic reports of Lincoln’s address even began to outpace reality; for instance, the Associated Press report falsely claimed it had been accompanied by “long continued applause.”

[Stupid Government Tricks]

A lot of the muted response was because speeches of the time routinely lasted for hours, so his 5 minute (?) talk was not recognized as a "real" speech.

[Roger S]

Or look at the coverage of Kamala Harris's dopey DNC speech. You might think that she said something of substance.

[TwelveInchPianist]

There's an interesting article that sheds some insight about how AI models can hallucinate stuff.

[Mr. Bumble]

"It seems lawyers are not the only ones who need to be careful relying on the veracity of AI-generated sources."
That would seem to be obvious.

What happens when AI generated information becomes more prevalent and gets scraped up along with non AI material to generate an output.

[Riva]

A problem. And even worse since the left is programming AI to generate their garbage.

[Life of Brian]

What happens when AI generated information becomes more prevalent and gets scraped up along with non AI material to generate an output.

Those who don't learn from entropy are doomed to repeat it.

[Stupid Government Tricks]

Pffff. AI can't begin to match politicians.

[Jokibif0]

"It insists upon itself." - Peter Griffin on The Godfather

[Carl_N_Brown]

AI mines text from whatever datasets of text it has access to, but AI lacks comprehension of the text it analyses nor does it have any human experience in the subjects of its artificial analysis.

[Roger S]

Does anyone expect Hollywood to be honest about summarizing reviews? All its movies are fictionalized, even when claiming to tell a true story. ChatGPT might have been more honest than some real human publicist.

[Stephen Lathrop]

When AI gets fully mobilized, the public life of the nation will be replaced, almost entirely, by AI-generated crap. The only natural persons still online will be dupes who suppose stupidly that actual people are paying attention to them.

Of course there is a remedy for that, but no one wants it. The remedy is to return to the practice of universal private editing prior to publication. Internet utopians hate that idea, but it will be the only realistic choice. Bring back private prior editing, or forget the notion of public life.

[David Nieporent]

"Here's a bad non-solution to a non-problem."