AUSTIN (KXAN) -- Results from a nationwide survey show a vast majority of respondents want to know when and how journalists are using artificial intelligence.
Eleven news organizations -- including the Associated Press, USA Today and KXAN -- collected more than 6,000 responses from their audiences in July and August. The organization Trusting News, which works to promote ways journalists can earn their audience's trust, compiled the responses and professional researchers analyzed them.
In addition to the 94% of respondents who want journalists to disclose AI use, 87% said they wanted to know why it was being used, while 94% wanted to know how journalists would ensure the work remains ethical and accurate.
Among the AI tools people felt most comfortable with journalists using were for checking spelling and grammar (79% comfortable or somewhat comfortable) and transcribing interviews (67% comfortable or somewhat comfortable). Those tools don't necessarily use generative AI, which is a form of AI that creates text or images when asked and has more recently come to the forefront of discussions as tools like ChatGPT became available for a wide range of people to use.
KXAN currently uses transcription and spellchecking tools with humans reviewing for accuracy, a practice that Trusting News said is broadly accepted industry-wide.
"There's a lot of ways newsrooms are using it to be more efficient," said Trusting News Assistant Director Lynn Walsh. "How can it help with my daily job?"
On the other hand, 76% of respondents said they were uncomfortable with using AI to serve as voiceovers or anchors, while 82% said they were uncomfortable with AI writing stories without human review.
KXAN does not do either of these things.
"If artificial intelligence is used to create content, generally speaking, the people we surveyed are not super comfortable with that," Walsh said.
Even when humans review AI suggestions, more than 51% said they were uncomfortable with AI writing headlines, a number that jumped to 55% when it comes to AI writing stories.
KXAN does not use AI to write stories but is experimenting with a tool called Nota that suggests headlines and summaries based on the stories our journalists write. Our journalists review the suggestions and may incorporate some aspects of them into our final headlines.
The First Warning Weather Team also uses a tool called PollenSense for its pollen count, which uses artificial intelligence to identify and count different types of pollen, allowing us to have more accurate numbers throughout the day.
Walsh said as AI becomes more widely adopted and accepted across all industries, it's important for newsrooms to prioritize accuracy, ethics and human review.
"It can help us be more efficient, which allows us to do more and produce better journalism for the people we serve, which is a win-win for everyone," Walsh said.
KXAN wants to be innovative in experimenting with new tools and technologies. It's important to understand how AI tools work because experiencing their limitations and benefits helps us discover potential ways to improve our workflows as well as educate our audience about them. However, we will not share anything publicly that was made using AI tools until the tool has been thoroughly vetted, and the output checked by one or more of our journalists.
Every story, whether an AI tool was involved or not, must always be reviewed by at least one other person before it is published on our site or makes its way on air.
KXAN is currently working on an AI policy we plan to post on our website. We will continue to discuss how best to disclose the use of any AI tools we decide to try. If you have feedback to share, email kate.winkle@kxan.com.
We will also work to cover how AI impacts our viewers. Send any story ideas you have to reportit@kxan.com.
