An overview of the launch of JournalismAI’s Generating Change Report

We recently hosted the launch of JournalismAI’s 2023 report: Generating Change: A global survey of what news organisations are doing with artificial intelligence. It was a pleasure to host the launch, and to discover and discuss the findings of the report with attendees from across the news and publishing industry. We were joined by over 130 people in-person, with another 1,300 watching the livestream.

In this article, we will highlight some of the key findings from the JournalismAI report, as well as showcase some of the takeaways from our panel discussion with:

• Dr Maryam Ahmed (BBC News, Senior Data Scientist/Journalist)
• Jane Barrett (Thomson Reuters, Global Editor for Media News Strategy)
• George Montagu (FT Strategies, Manager & Head of Insights)
• moderated by Professor Charlie Beckett

JournalismAI is a global initiative that empowers news organisations to use artificial intelligence responsibly. It is a project of Polis – the LSE's journalism think-tank – and is supported by the Google News Initiative. They aim to educate and inform newsrooms on AI and the opportunities and risks that come with it. Their research will foster debate about the editorial, ethical, and financial implications of using AI in journalism, and to help news organisations chart their way forward.

This is the second report from JournalismAI: their 2019 report surveyed 71 organisations from 32 countries. In 2019, JournalismAI showed that, whilst its role in the newsroom is uncertain, AI is an area that all news organisations should be paying attention to. In the years since then, ‘traditional’ (non-generative) AI use cases have become relatively tried-and-tested.

The 2023 Generating Change report builds upon these findings by surveying a larger and more diverse group from the industry, and considering the potential impacts of generative AI (GenAI).

Prior to GenAI, the essential abilities of AI algorithms were predominantly centred on processing unstructured data and performing regression (scoring), classification and clustering. However, with the dramatic rise of GenAI, most commonly observed through the release of consumer-friendly products like ChatGPT, Bard and Midjourney, a new set of opportunities and challenges have arisen for the news industry. JournalismAI’s report explores what is happening with AI in the news industry and provides guidance on how to navigate these changes. Read FT Strategies’ analysis of AI functionalities and implications for the publishing industry here: Artificial Intelligence 101: Core functionalities and foundational use cases for digital publishers.

The 2023 report included 105 organisations spread over 46 countries. This research was made up of insights from organisations both large and small, nonprofit and for-profit, and from news, magazines and broadcast organisations. The research was conducted through surveys and interviews from April and March 2023. The report was supplemented by interviews and conversations at conferences.

Below are the topics that were covered at the event:

• 85% of newsrooms currently experiment with GenAI
• GenAI presents opportunity, but more skills are required
• A lack of understanding and strategy is still present across newsrooms
• JournalismAI recommend steps to approach AI strategy
• Panel discussion

The report shows that most (85%) newsrooms have experimented with GenAI and newsrooms are already using AI or GenAI across parts of their content creation process, as shown below:

• 75% - News gathering
• 90% - News production
• 80% - News distribution

Newsgathering

AI can be used in the news gathering process in many ways, with the most common use cases being data processing and analysis.

Newsrooms are exploring tools that automate transcription, extract text from images, and structure data after gathering:

• Optical Character Recognition (OCR)
• Speech-to-text
• Text Extraction

Newsrooms are also looking at tools that can sift through large amounts of data and detect patterns:

• Trend Detection
• News Discovery

News Production

News production is the area where most newsrooms are currently using AI to enhance and augment their journalism, with 90% of newsrooms reporting this in the 2023 report. Newsrooms report multiple ways of using AI in the news production phase, including:

• Helping journalists fact-check against the publisher's updated fact database
• Summarising content
• Writing headlines
• Producing targeted newsletters
• Code writing
• Source assessments
• Visual storytelling

News Distribution

In this phase of the content cycle, newsrooms have found the widest range of use for AI (which can be used to optimise both reader value and engagement), including:

• More accurately matching content to targeted audiences at scale
• Optimising their content mediums (i.e. making an article audible using text-to-speech)
• Social media distribution optimisation
• Search Engine Optimisation (SEO)

Almost three-quarters of organisations in the report agreed that GenAI presents new opportunities for their newsrooms, an expected new finding given the proliferation of GenAI since the 2019 survey. Most (around 80%) respondents see themselves using AI more in their newsrooms in the four following areas:

1. Fact-checking and disinformation analysis
2. Content personalisation and automation
3. Text summarization and generation
4. Using chatbots to conduct preliminary interviews and gauge public sentiment on issues

To capitalise on these opportunities, newsrooms are looking at upskilling current staff and bringing in staff with the skills required. Most (90%) newsrooms will look to train their current journalists on ‘AI-augmented journalism’ skills, including AI literacy, data analysis, and ethical considerations. At the same time, around 25% of respondents are looking to bring in AI specialists, including, AI managers, data scientists, and developers.

The report notes the need for cross-departmental collaboration to make the most of AI’s opportunities.

As found in the 2019 report, and as still remains a challenge today, newsrooms are struggling with a lack of knowledge and skills to understand AI and the opportunities and threats that come with it. This lack of understanding is shown in the report, as 50% of newsrooms don’t have an AI strategy, and a further 17% are unclear about their strategy. 40% of newsrooms believe AI poses a challenge to their business, with a further 52% unsure if, and what, challenges it poses. On top of this, 40% of newsrooms haven’t changed their approach to AI since 2019, despite the development and access of GenAI over the last year.

Of the newsrooms that are concerned about AI, their worries are that AI and GenAI will exacerbate misinformation, bias, and fake news.

In regard to the practicality of implementing AI into newsrooms, the report shows that the most challenging areas for AI integration in the newsroom are:

• Technological challenges & financial constraints
• Ethical concerns
• Cultural challenges
• Managerial challenges

JournalismAI recommended a set of strategic next steps for newsrooms thinking about what to do and where to start with AI:

1. Get informed (to build common understanding)
2. Broaden AI literacy (to instil agency)
3. Assign responsibility
4. Test, iterate, repeat
5. Create guidelines
6. Collaborate and network

We were also honoured to be joined by three incredible guests in our panel discussion, moderated by Professor Charlie Beckett of JournalismAI:

• Dr Maryam Ahmed - BBC News - Senior Data Scientist and Journalist
• Jane Barrett - Thomson Reuters - Global Editor for Media News Strategy
• George Montagu - FT Strategies - Manager and Head of Insights

AI and GenAI have seen a lot of development, investment, and study over the last year, especially since the launch of ChatGPT at the end of 2022. Our panel had a lively discussion over three areas:

• Where AI & GenAI is and isn’t being used in journalism:

Firstly, Dr Maryam Ahmed explained how approximately 70% of her work does not involve AI at all, but instead relies on a toolbox of other data analysis techniques. In the other 30%, AI is used as a tool to make work more efficient, including using optical character recognition (OCR) to make sense of messy text and documents, and to speed up, and help with more advanced data analysis and “fancy statistics”. Another interesting area for Dr Maryam is flipping the question on its head and trying to understand the inner workings of AI and the algorithms themselves. “As AI is becoming more important in our daily lives, it’s important to understand how it works, uncovering the systems and showing where the bias or blind spots lay.”

• How organisations as a whole can start experimenting with AI & GenAI:

Jane Barrett explained that Reuters are approaching AI from three main angles:

1. Guidelines - setting the foundation of how the business can think about using AI and the tone for the rest of the organisation and all of the experiments that follow.
2. Experimenting and playing with AI - in cross-disciplinary teams, from journalists, marketing, and business strategy, Reuters are working together internally to experiment with how they can make the most out of AI.
3. Implementation and integration - understanding what it will take to train journalists, which parts should sit with Editorial versus with Technology, how to track AI usage, and to what extent should AI workflows be integrated with both in-house and third-party tools.

• How organisations should view AI & GenAI.

George Montagu explained how he’s approaching AI from three perspectives:

1. Education - George and the team at FT Strategies are working collaboratively with other departments across the Financial Times to understand how leading news organisations are thinking about and using AI.
2. Practical implementation - While GenAI is still being explored, there are many established use cases of non-generative AI that can be applied to businesses in the news industry today. Some of these use cases include propensity modelling, auto-tagging archive content, advanced user segmentation, and more.
3. Understanding GenAI - George is also speaking to people in the wider news and tech industries. He’s found it interesting that the common thread through his conversations has been “What we could do with AI” rather than putting the reader first and asking “What are the reader's needs that could be solved with AI”.

GenAI has presented new opportunities and challenges for newsrooms of all shapes and sizes. While it can be an overwhelming topic that is tempting to avoid, it is important for business sustainability that organisations recognise a need for training and upskilling. Read how our panel discussed Generative AI’s potential:

• “It’s absolutely not going to replace the work or expertise of an investigative journalist.” Whilst there are lots of exciting use cases for how AI and GenAI can augment and aid the work of an investigative journalist, our panel mentioned that it won’t be able to replace those discovering and interrogating the key stories.
• Areas it can be used - As previously mentioned, we’ve already been able to identify some exciting areas that AI & GenAI can help journalists produce better, or previously unfeasible work. This can be anything from making sense of large amounts of unstructured data, as Dr Maryam Ahmed is currently experimenting with, or being able to aid investigations that wouldn’t be feasible for one-person or small newsrooms to currently conduct, for example, using GenAI to simulate 1000s of interactions to investigate whether estate agents would treat a 35-year-old male executive the same as a 25-year-old single mother claiming benefits.
• What’s holding it back? - Our panel agreed that at the moment, the errors that GenAI is prone to making are stopping it from being a reliable partner in most use cases. When using GenAI currently, there is a lot of verification and validation that has to be manually done before being able to progress with the information produced.

News organisations have been forced to change multiple times over the last two decades, from adapting their revenue models to the digital world, social media personalities and profiles breaking news faster than traditional news organisations, and bloggers and YouTube opening the world of journalism up to everyone. Now, the news industry is on the brink of another drastic change, AI & GenAI. Read how our panel discussed how their organisations are adapting to the latest change:

• “This time it feels different”. Jane Barrett discussed how she’s managed previous transformational shifts in the news industry, but that reactions to change are different this time. People are recognising AI as an unavoidable change which is very much underway, and so feel that Jane is “helping me through something that’s being done to all of us”. This mindset shift is making it easier for organisations to come together and move forward as one when approaching AI.
• “Gen AI will be a bigger change than the internet”. There is natural hesitation when it comes to making big changes to organisations and in newsrooms, especially when standing in front of a technology that could have such wide-reaching implications. At the same time, there are also people who are very enthusiastic about AI and want to get started immediately. Our panel found that this is where collaboration and having guidelines in place can help an organisation plan their way forward together.
• Where are the naysayers? The panel discussed how in previous changes to the news industry there have been people that wanted to push back against, and not accept, the change that was in front of them. But with AI, everyone feels in agreement that this is something that can’t be avoided or ignored. Again, helping organisations make positive action towards AI changes.

As we’re at the beginning of the AI ‘revolution’, organisations are looking for thought leaders to help them understand and navigate the approaching changes.

• “Test, Iterate, Repeat” - Our panel discussed how this is a great way to think about bringing AI into your organisation. At the moment, AI is a long way from changing the revenue model of the majority of organisations, so taking time to test, iterate, and repeat what’s working is how organisations and newsrooms will be able to introduce AI into their workflow, business processes, and output.
• Use AI strategically - Ensuring that AI is being utilised to support your existing business goals (such as an organisational North Star) is an important way to guide experimentation and implementation efforts.
• Align your AI use with your values - In addition to aligning your use of AI with your organisation's strategy, it’s vital to consider your organisation's values and how AI can support, enhance, and help you live up to these values. Having your organisation's values at heart will help you decide which opportunities to pursue and which risks to mitigate against.

Register below to recieve a copy of the report, and watch the full event back on-demand.