April 20, 2024
A.I

Tips and tactics for 2024

Leaders in using generative AI in newsrooms shared tips, tools and some optimistic predictions ahead of the holidays at NCTJ’s AI in Journalism event.

To start the new year, Press Gazette has rounded up the most useful lessons shared, as well as some thoughts on what’s around the corner.

Read on to learn more about how Reuters has addressed the risk of AI data leaks, how to build your own AI co-pilot, and why a Sun staffer believes investing in generative AI should pay off for newsrooms. “imminently.”

Reuters experiment with ChatGPT keeping it within ‘a good, strong fence’

Several editors have previously expressed caution about embedding their copy on ChatGPT. The Telegraph, for example, banned staff from using the tool to edit texts, warning that the data breach could breach both the publisher’s copyright and data protection laws.

Jane Barrett, global media news strategy editor at Reuters, said at the event that the news agency has tried to address this issue by putting “a strong fence around” the generative AI experimentation it has done.


“We have our own ChatGPT instance through Microsoft Azure, which allows us to play safely,” he said.

“And I think that’s something important: to experiment, but to experiment responsibly. He plays, but he plays with confidence. That was the first big goal we learned.”

Content from our partners

“Playing” was the topic of Barrett’s conversation with Sky News presenter Vanessa Baffoe. She said: “Everyone should do it just to play. Find your inner child, use some of these tools…

“I said this at a conference a while ago, and someone added a really important thing, which I’m stealing a lot from: play and ask about things you know a lot about. Because then you will understand its limitations and you will understand where it goes wrong.”

One limit he pointed out that was particularly relevant to journalists was ChatGPT’s understanding of quotes: “The AI ​​doesn’t know what a quote is… If you’re summarizing a story, you want to keep the quote separate, you don’t want to summarize “. the quote as if it were part of the story… We are now working with our data scientists to determine: how can we teach the machine what a quote really is and leave them alone?

Newsquest Editorial AI Head Creates AI Assistant for Journalists Live on Stage

Jody Doherty-Smith, AI editor-in-chief at Newsquest, said that because generative AI tools will improve and change at a rapid pace, “the most enduring skill is knowing how to talk to AI.”

During the talk he revealed that in six months Newsquest had gone from hiring its first AI-assisted reporter to employing seven across the country.

[Read more: How Newsquest and its seven AI-assisted reporters are using ChatGPT]

A slide from a presentation given by Newsquest's AI Editor-in-Chief, Jody Doherty-Cove.  She says
A slide from a presentation given by Newsquest’s AI Editor-in-Chief, Jody Doherty-Cove. She says that “AI agents will change everything,” before saying in a bulleted list that “digital assistants can perform a variety of tasks, from in-depth investigations to data analysis and preliminary reports”; that “journalists can go deeper, using traditional human intelligence to craft narratives”; that it is not “just about delegating tasks, but about enhancing your skills” and that “although it is built using GPT, it goes much further than ChatGPT. Once built [AI agents] “They will continue to fulfill their purpose.” Then, the image of a cute robot is marked with arrows, suggesting that an AI agent has three components: its knowledge base, the actions it has been able to perform, and its purpose, that is, the job that a journalist needs you to complete. Photo: Press Gazette

Doherty-Smith said journalists would benefit most from AI by identifying the specific problems they need to solve and setting up specialized “agents” (custom AI models) to address them.

Generative AI chatbots can alter their output based on cues given to them by a user. Doherty-Smith showed attendees a table showing the types of prompts that can be given to an AI, including requests to restructure or reformat given information, to obtain information from elsewhere, or to brainstorm more information on a given topic.

He then attempted live on stage to use ChatGPT’s GPT Builder feature to create an internet-linked agent that could write and submit freedom of information requests. After entering a short series of prompts, she managed to create an agent that could format the user’s questions into a full FOI request and that could retrieve FOI email addresses from local council websites.

He did not perform the more technically complicated task of giving you the ability to send an email, but said it was possible and that Newsquest reporters had already done it.

He said that “one of the things, actually, when we were playing with this and we first linked it to stocks [the ability to interact with the web] – sent one [an FOI request] without them telling you.” Accordingly, he added that it was important to tell the agent to present any draft email to the journalist before sending it.

“Remember that [while] experimenting is very fun, publishing is very serious,” he said. “So make sure you have safeguards in place. “We have a code of conduct at Newsquest… to make sure everything we publish is correct.”

But he concluded: “AI agents are here. so move fast and get your own.”

A typical freedom of information request, but written by an AI as directed by Newsquest's chief editorial AI, Jody Doherty-Cove.
A typical freedom of information request, but written by an AI as directed by Newsquest’s chief editorial AI, Jody Doherty-Cove. Photo: Press Gazette

Bullish predictions for generative AI and journalism in the short term

While news industry predictions about the effects of artificial intelligence often emphasize future risks, there were some notably optimistic comments at the NCTJ event.

For example, when Joanna Webster, acting video and photo editor at Thomson Reuters, asked when media outlets could expect to see “a return on this investment” in AI, Nadine Forshaw of The Sun said it should be “imminently”.

“It really depends on what your proof of concepts is and what you’re doing,” he said.

“If you play with him for fun, great. [but] it may not bear fruit right away. If you’re looking for things that are really going to save journalists time, that are going to allow you to reuse content, whatever it is, I think as long as what you’re creating fits the problems that you have in your “As it happens.” build news, there’s no reason why the payoff shouldn’t be pretty immediate.”

Similarly, Barrett said that at Reuters “we fully anticipate a time when there will be some stories that can be published without a human being monitoring each story. But at that point we must make it very, very, very clear that this story has been made by AI.

“We currently publish some translations, just for our financial clients, I would say, because they simply wanted more volume. [and] They weren’t going to pay us to have people translate things. So we’re testing some machine translations that aren’t reviewed by a human before being published, but that say at the top in big letters: ‘This was translated by a machine.'”

Barrett added that Reuters already employs staff whose full-time job is to carry out spot checks “a proportion of them every day.”

Manjiri Kulkarni Carey, editor of the BBC’s innovation arm, BBC News Labs, said the adoption of generative AI in newsrooms will change the skills editors look for in staff.

“I consider myself a kind of ‘bridge’ producer, because I can talk to software engineers, but I’m also a journalist, and there just aren’t enough people like that.

“Students who are getting ahead will need to do this more and more… If you can develop those basic skills, you can do anything. But the more skills you have and the more you can perform these types of bridging roles, the better.

“Because newsrooms are only going to evolve in that direction… they’re going to say, ‘How are we going to write a program that will interrogate these 15,000 pages of data to publish five stories?’

“You need to be able to distill, interpret and work with product people, engineering people and all those technology people to get those stories.”

Gavin Allen, professor of journalism and digital at Cardiff University’s School of Journalism, Media and Culture, said on that topic that his course had begun teaching students how to use rapid AI engineering for their journalism.

More broadly, he said: “AI is, perhaps, an opportunity to make some money in this industry.”

Bonus: Click to access the list of journalism and artificial intelligence case studies from the Google News Initiative.

Email pged@pressgazette.co.uk to point out errors, provide story tips, or submit a letter for publication on our blog “Letters Page”

Leave a Reply

Your email address will not be published. Required fields are marked *