Destroyer or I


Should we panic about the rise of AI? Stephanie Power says it’s all about how and why media companies use it


have been working on stories about the impact of AI on our lives for some time. It’s always so much easier to focus on the AI takeover of law or medicine, or even of our relationships. But scrolls through LinkedIn


have drawn my attention to the rise of AI in journalism. In the summer, The Washington Post advertised a new role, director of AI and machine learning, guiding the development of ‘intelligent, ethical, and secure platforms that power everything from operational efficiency to innovative newsroom tools’. The job ad says that the successful candidate will believe deeply in the mission of journalism. The salary range isn’t bad at between $200,000 and $350,000. But you need a master’s degree in computer science, machine learning or statistics to apply. They’d prefer a PhD. I’m not sure there are too many journalists with PhDs in computer science. I’d also spotted someone brandishing a trophy from the


Association of Online Publishers. The award – Best Digital Publishing Innovation. The winners – Newsquest, for their use of AI assisted reporters.


Newsquest argues this


innovation means reporters can spend more time in local communities. I thought it might be fun to spend a morning with one of these new reporters for this piece but, unfortunately, I couldn’t get anyone to engage with me. The head of AI didn’t respond to my LinkedIn messages and neither did Toby Granville, the director of editorial development. Why don’t they want to showcase their award-winning


work? In March, Granville spoke to the Society of Editors Media Freedom Conference, telling the audience the company now had 36 AI assisted reporters, a move that “frees up the rest of the newsroom to actually be out pounding the beat, knocking on doors, getting original content that people are willing to pay for”. Granville was asked by Press Gazette how AI reporters


would be able to progress professionally without traditional beat reporting. He suggested they were


learning a different set of skills, having regular AI literacy training and brainstorming new ideas for using AI in journalism. He said these reporters could go back to a traditional role if they wanted. Everyone is experimenting. This summer, the BBC


released new editorial guidelines. Their section on AI states that its use must:


• Never undermine audience trust • Always be transparent and accountable with effective and informed human oversight • Always be used consistently with BBC editorial values, in


Sources, walls and energy: a user’s guide


There are many ways you can use AI to help you in your work. I have used it to double-check facts, help shape up workshop sessions for media training and write those cringey biographical blurbs you are sometimes asked to do. First, my advice would be to


ask yourself who your commissioning editor is. Does their organisation request transparency? Do you need to ask/tell them how you are


16 | theJournalist


using AI and which large language model (LLM)? Second, use an LLM that


provides detailed sources. I use Perplexity Pro, which in turn uses several other models, including xAI, Anthropic, Open AI and Google. But just because there is a source attached to a piece of information, that is not a get-out-of-jail-free card. Some sources are better than others. So check and check again.


Some in-house LLMs will


act like a walled garden so any information you put in should be protected. Off-the-shelf LLMs won’t. So be careful about your data. I wouldn’t have a problem asking Perplexity for help with a media training exercise. But I won’t be putting my potential soon-to-be best-selling novel in any time soon. The results you get are only


as good as the prompts you put in. So think through what


you are asking for. Be clear about what you want. Try to get the question right first time because advanced LLMs use a lot of computational power. Even MIT says it’s nearly impossible to work out how much power your questions might use. It suggests asking ChatGPT 15 questions about fundraising, playing round with


10 flyer designs, and three attempts at a five second promotional video would be the equivalent of running a microwave for more than three hours. Small questions will make you feel less guilty.


Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26  |  Page 27  |  Page 28