Comedian: The rise of AI comedy reviews 'makes me sick' as ChatGPT used at Fringe
Performers and producers say they have seen reviews they are certain are generated by AI
Some Fringe reviews are being generated by AI, performers have warned.
Comedian Kieran Hodgson said he believed some online reviewers working at the Fringe were recording shows and uploading them to services such as ChatGPT to generate an automatic review, saying the practice "makes me sick".
Speaking at an industry panel event, Mr Hodgson said: "This year particularly, we have noticed AI reviewing coming in and it makes me sick. It's very clear what happens: someone sits there, presses record and at the end of it, they have generated a review."
He said the rise of AI reviewing highlighted the need for legitimate and rigorous awards schemes in the comedy sector.
He said: "There's stars flying around and words flying around and while everyone loves and needs good reviews and good stars, there is an integrity about how the [awards] panel does it. It's not a fix and it's nice to know it is taken seriously."
Another producer, who asked not to be named, said one review of a show written for a website had mentioned the performance of a song at the end of the act - which a human reviewer would have known was not a song, but could have been mistaken for such through words spoken by the performer over the music.
"The only possible explanation is that a recording of the show, which has the music playing, has been uploaded to AI," he said.
Paul Levy, editor and founder of reviews website the FringeReview, said his publication has strict rules surrounding the use of AI, but warned the practice was "widespread" in the sector.
He said it is known that some reviewers working for websites are uploading recordings to generate automatic reviews.
He said: "That is exactly what they are doing. We know it is widespread but we don't have the evidence because it is not 100 per cent possible to prove it."
Mr Levy, who also works as a senior lecturer in the School of Business and Law at Brighton University, where he teaches AI ethics, said online tools to check whether something has been written with AI are not reliable.
"The holy grail is the search for a foolproof tool, but it's not there yet,” he said.
"One of the telltale signs that AI is likely to have been used is the use of the em dash, the longer hyphen, but the problem is that use of the em dash is quite widespread in academia for example, so a reviewer from that background may well be using it naturally."
The Fringe Society said it was not aware of any AI reviews on artists’ show listing on its website and said it has monitoring systems in place for professional reviews on its site.
A spokesperson said: “Many industries are now engaging with AI tools across a range of areas within the workplace. Some media outlets may have their own rules or standards around how AI is used when writing reviews.
“We would advise artists to be wary of reviews that come from media they don't recognise from their audience and/or see on their ticketing reports. If artists become aware of questionable practices during the Fringe, we would encourage them to reach out for support from our Artist Services team.”
Comments
Want to join the conversation? Please or to comment on this article.