How are publishers getting the most out of AI? And what challenges still need to be addressed with this new technology? A recent Civic Journalism Lab event looked to tackle both the benefits its bringing to the industry and what concerns still need to be tackled, with the panel formed of Aimee Rinehart, senior product manager for AI strategy at The Associated Press, Felix Simon, research fellow in AI and news at the Reuters Institute for the Study of Journalism, Sophie Smith Galer, digital journalist and content creator, and Chris Stokel-Walker, freelance journalist for titles including The Times, Wired and The Guardian.

AI in practice right now

Media organisations and journalists are using AI tools more frequently now, and for Aimee Rinehart there has been one big development:

‘Transcription is a game changer. It just makes the job so much easier and it solves a need. Another good use of AI is information retrieval from existing archives. We saw that with the Panama Papers years ago – they wouldn’t have been able to trove through all of those documents without some assistance.’

Sophia Smith Galer, who has created her own AI tool in Sophina, spoke of the benefits that Fyxer AI has brought her as a freelance journalist: ‘It’s transformed how I do emails. The amount of time it has sliced off my email responding time, which takes up a lot of my job, has been really helpful.’

Felix Simon spoke about early AI use cases at the Financial Times which helped flag content that had too much bias in one direction, for example gender representation. He also mentioned a Finnish newspaper that has built a tool which makes suggestions based on that article as to what else might be interesting or insightful to include for the reader – for example on a financial story, having a text box explaining the key terms.

The need for transparency

All publishers will have editorial checks or guidelines in place for their journalists, or adhere to rules set out by bodies such as IPSO. While some publishers have done this for their use of AI, there is no universal agreement on what should and should not be included. Felix, from his work at the Reuters Institute, has noticed a unifying point:

‘Something that unites most of the guidelines that I see at this point is to focus on being transparent about AI use, especially where there’s a risk that audiences could feel misled from the use.’

He went onto talk about what audiences themselves wanted and it was both the purpose of the news organisation for using AI, and also what benefits were in it for them. Felix said the audience wanted to know ‘how does it make the journalism they receive better, more interesting, more qualitative and more personalised.’

Aimee said that at the AP they are already adding taglines at the bottom of articles when they use AI to show the public their ethics and standards and for transparency. Chris also agreed that the use of AI needs to be disclosed, whether in the reporting process or the production process and added:

‘In an uncertain information environment where fake news is bandied about a lot, I think AI just amplifies that risk even more. It means we have to be cleaner than clean.’

Other considerations and possibilities when using AI

When developing her tool Sophina, Sophia has had to consider putting guardrails in place so that it’s protected and only being used by those with good intentions but it requires two things:

‘If you really want to set up lots of guardrails for protecting the tool from being abused, you need to know more about the users. They need to give more information, they can’t be anonymous people that you can’t tell what they’re doing. The other thing is the amount of money. The sum that has to go into the development to establish all of these guardrails is enormous.’

Before spending the money to put these in place, another consideration is whether the AI tool you have developed or bought is actually accomplishing a need for the company or for journalists. Felix explained ‘we have to put proper processes in place and lots of organisations need to assess, does it actually make a difference? I think that’s something which is slowly being talked about more and more in the news industry, but it’s still not quite where it needs to be.’

The potential for ways in which AI could help the media are vast. Aimee sees potential in reformatting articles to ‘meet the audiences where they are’. She said ‘a lot of people don’t have time to sit down and read (an article) but could absolutely have it told to them. It therefore holds enormous promise for that multi-platform distribution of a basic article.’