‘The arrival of Generative AI is going to be a bigger change than the arrival of the internet’

It's not a question of whether we experiment with AI but which AI tools can best support journalists or newsrooms to reach their goals and how we integrate them

Sep 27, 2023

Hello and welcome to the second edition of the Happy Journalist… Before I start, I just wanted to say a big thank you to all of you who have signed up in the last week - I’ve been blown away by your support. If you haven’t yet signed up, please do hit the subscribe button below to get a free copy of the newsletter in your inbox every Wednesday.

I was lucky enough to attend Google’s Newsgeist in Italy this weekend to talk through some of the biggest issues facing journalism today. It was no surprise that the main topic of conversation was AI. In fact at one point I went for a swim in Lake Maggiore to find myself surrounded by six different groups of journalists all talking about AI!

The ‘unconference’ (delegates have to decide the conference agenda when they arrive) covered a lot of ground from impact to solutions journalism, trust to filter bubbles, revenue models to climate - much of which I’ll return to in the coming weeks.

But as the event came days after the launch of the JournalismAI report and it’s undoubtedly the topic of 2023, it seems pertinent to start there.

The panel at the launch of the Journalism AI report at FT Strategies.

The report, by journalism’s leading light on AI, Professor Charlie Beckett and Mira Yaseen, from the London School of Economics, revealed that 85% of respondents had experimented with Gen AI and 75% of newsrooms were already using AI in some form in their news gathering, production or distribution processes.

It outlined the challenges - technical, ethical and cultural - in integrating AI into newsrooms; the current use cases for generative AI; and the potential opportunities it could create for journalists. And it set out the next steps for news organisations to take in setting out an AI strategy:

1 Get informed.

2 Broaden AI literacy.

3 Assign responsibility.

4 Test, iterate, repeat.

5 Draw up guidelines.

6 Collaborate and network.

I would add a 7th step to the list - to work out how AI tools meet your audience needs, or solve a problem you have and/ or support the goals and values of your business. It’s futile to try to integrate tools into a newsroom that aren’t going to help you reach a bigger audience or build better relationships with your audience or support your revenue goals.

Far from threatening the jobs of journalists, the clear message from the report was just how much AI has the potential to benefit and change journalism.

At Newsgeist, many journalists had examples of AI use cases that had worked for them, particularly in transcription, translation, data cleaning and summarisation. But much of the focus was on the flaws inherent in current models and the dangers it posed if used in the wrong way.

For me, the week’s AI discussions have posed three questions:

📝 What do we (journalists, newsrooms) actually want to achieve with AI? And how can it help journalists and media organisations reach their goals?

📝 Are we focusing too much on the flaws and hallucinations of AI because we’re scared that if it was flawless it would be too dangerous?

📝 And when Generative AI gets really good, how do we differentiate human content from AI? In five years’ time, will the public care who or what is writing something if they’re getting good journalism?

It’s not a question of whether or not to learn, test and iterate with AI, but which tools to start with. Because there’s little doubt that AI will change everything about the way we work in the next few years.

As Jane Barrett, Global Editor of Reuters, told the audience at the Journalism AI report launch at the FT last week: “There is a general awareness that generative AI is going to change a lot of how we live and work. [Thomson Reuters CEO] Steve Hasker actually said that he thinks the arrival of generative AI is going to be a bigger change than the arrival of the internet.”

I think he’s right.

AI experiment of the week

Tool: HeyGen

Script: Hi, I'm Jacqui Merrington and I'm testing HeyGen - a new AI product that makes scripts and pictures into avatar-led videos. This was just a free trial and look what it can do! Testing tools like these is part of my journey to establish the ways in which AI can actually benefit journalism and not just make it more efficient. Turning text into video is something that many newsrooms don't have the capacity to do and while a talking photo is not the answer every time, it can enable a journalist to produce an explainer in superquick time.

Response:

Tips: HeyGen is an AI video tool that can turn a still image and a script into a video in seconds. I uploaded a headshot, wrote a quick script and then chose from a selection of voices - they’re available in dozens of different languages. The first video is free - after that you pay from $48 a month for 30 credits.