– The use of technology is not the problem, but a humorist who has abused the journalist’s trust to promote his podcast, writes Eirin Larsen about Harald Eia. Photo: Thor Brødreskift / Nordiske Mediedager / Wikimedia Commons / Kristine Sterud
COMMENT:
“Regardless of Chat GPT, it is the person, not the technology, who is responsible. Therefore, Dagbladet should not have published Harald Eia’s submitted hesitation”, writes Eirin Larsen.
Artificial intelligence is buzzing on all channels, not least in Tore Sagen and Harald Eia’s podcast, where especially the AI-generated realistic voice of Jonas Gahr Støre has received a lot of attention. The presenters play with technology and provide entertaining public information about opportunities and dangers.
This week revealed Aftenposten that Harald Eia had left Chat GPT to answer in Dagbladet’s book column, which Eia confirmed in his podcast.
Dagbladet unpublished the interview following Aftenposten’s suspicions. The news editor emphasized that the journalist checked that the authors and book titles Eia referred to exist.
The answers were fact-checked, but they remained really, really strange.
In January, NRK Sørlandet was deceived by a PR agency that sent in a realistic AI-generated image that they claimed was a picture of “Trude Ailin Winther”. To TV 2 it is stated that the motivation was to test the control that is done with reader images. The editor apologized and later the NRK office closed the inbox for viewer images. Kind of bankrupt. Checking all photos costs too much.
The stunts can be compared, but there is an important essential difference between images and text generated with generative artificial intelligence. While realistic images appear to represent reality, text purports to represent the opinions of the sender.
One of the ethical dilemmas surrounding generative artificial intelligence in Norwegian newsrooms is whether it should be allowed to submit debate posts written with Chat GPT. The most obvious argument against is of course that the text is not man-made, but it is not that simple.
Is all the text printed in the newspaper columns written by the person who sends it in today?
We use a spell checker, get help from a friend or a communication advisor. The criterion is quality, and the columns are reserved for people who can write or have the resources to get help.
As a result, many people are excluded from the social debate. There are many people today who want to write about the situation in Gaza or Iran and who do not have access to the means the newspapers require.
With the help of generative artificial intelligence, one can read in one’s own stories and opinions and get help with reformulation. You can make a draft of a reader’s post, which you revise and adjust until it says what you mean. Isn’t that okay?
You obviously run the risk of a bad text, but exactly that is reasonably technology-independent. Generative artificial intelligence is only a good tool if you master it, which also applies in the education sector.
In the future, editors and journalists will have to check and verify facts more often and better than today. The technology is developing, and it will become impossible to know whether text is AI-generated. The arms race between technology to create and technology to reveal is already underway.
Journalists must check that authors and books exist, as Dagbladet did. Maybe the debate moderators have to search and call all new submitters to check that they exist. If the gain is that more people can participate in the social debate, it is probably worth it.
Harald Eia confirms AI-generated interview
The interview is still unpublished, but can of course be republished with a note that Harald Eia chose to answer Dagbladet’s inquiry with KI-generated text. Then Eia is held accountable, and more people get the opportunity to discuss and learn. As Eia herself says in her own podcast: “Now Dagbladet has removed it, so now I don’t have to stand for anything.”
Should the text be rejected, low quality should be the reason. In many circles, “AI-generated” is already used as slang for something bad, regardless of technological origin. Criticizing everyman’s bad answers is of course also easier than arresting one of the country’s funniest people.
The fact that the journalist got a suspicion that Eia was playing pranks would of course also be grounds for a refund. The use of technology is not the problem, but a humorist who has abused the journalist’s trust to promote his podcast.
Harald Eia said sorry, but whatever: Professional players using the press as their personal PR apparatus is nothing new. But it’s not about technology either.
———————————————-
A slightly different version of this post has been published at Dagens Næringsliv. This version is reproduced with permission.
This is a comment, and expresses the writer’s opinion. Would you like to write for Medier24? Send your post to [email protected].