bbb

Collab

A global collaboration to experiment with AI

The JournalismAI Collab is a collaborative experiment launched in June 2020 as a platform for news organisations to come together and explore innovative solutions to improve their journalism through AI.

About 40 participants from 20+ news organisations worldwide focus on specific challenges that they selected and that they thought AI might help address. Together they imagine and test new ideas that might lead to developing new tools and inform future experiments.

Participants are organised in five international teams to explore the following questions:

How might we leverage AI to understand, identify and mitigate newsroom biases?

To better serve our societies, newsrooms need – and most times want – to do better on diversity and inclusion, both in terms of who works in news organisations and in what/how they report on the world. In the context of the Collab, the team is exploring how AI might be a resource for news organisations and the industry as a whole. The team is looking at this challenge in three different steps:

1. They are looking into where and how bias might manifest itself in newsrooms, focusing primarily on gender, racial and age bias.

2. They are exploring how AI might help identify bias, through experimenting with AI solutions that will help generate insights about binary gender representations.

3. And they are researching how AI tools and insights can practically contribute to newsroom diversity and inclusion goals.

The team will share their learnings through a website where they will also showcase the results of their practical AI experiments with assessing the gender representations of their respective news sites. They are approaching AI as a potential force for good in the media industry and trying to highlight how the same technologies people are scared of, might be used for good.

THE TEAM:

  • Agnes Stenbom, Responsible Data & AI Specialist, Schibsted (Sweden)
  • Aurore Malval, Digital Reporter, Nice-Matin (France)
  • Delfina Arambillet, Data Journalist, La Nación (Argentina)
  • Flor Coelho, New Media Research & Training Manager, La Nación (Argentina)
  • Issei Mori, Computer Science Intern, Nikkei (Japan)
  • Jane Barrett, Global Editor for Media News Strategy, Reuters (UK)
  • Michaëla Cancela-Kieffer, Deputy News Editor, AFP (France)
  • Paul Gallagher, Executive Editor, Digital Development, Reach plc (UK)
  • Ruth Kühn, Senior Technology Manager, Deutsche Welle (Germany)
  • Yosuke Suzuki, General Manager, Digital Business Development, Nikkei (Japan)

How might we create an automated suggestion engine that puts the power of our archives in the hands of our journalists?

The team wants to facilitate the creation of a suggestion engine that recommends articles from a newsroom’s archive to a reporter who is in the process of writing a story or is planning to write one. This would help journalists leverage existing knowledge to enrich new articles and add useful context and background information, thus showing how a story has evolved over time.

The goal of the team is to create a step-by-step menu that any newsroom can follow to implement a similar solution onto their existing CMS and tech stack, with the only prerequisite being the access to a suitably-sized content archive.

THE TEAM:

  • Flor Coelho, New Media Research & Training Manager, La Nación (Argentina)
  • Kristoffer Hecquet, Head of Development, Altinget (Denmark)
  • Momi Peralta, Data Project Manager, La Nación (Argentina)
  • Nick Cameron, Head of Performance, Archant (UK)
  • Sophie Casals, Solution Journalist, Nice-Matin (France)
  • David Corral, Head of Innovation, RTVE (Spain)
  • Melissa Stevens, Digital Editor, South China Morning Post (Hong Kong)
  • Padraic Cassidy, Editor, Automation & News Technology, Reuters (USA)
  • Sebastian Maulbeck, Senior Product Owner, Content Intelligence, Axel Springer (Germany)
  • Tim Nonner, Chief Data Scientist, TX Group (Switzerland)

How might we use methods of structured journalism to identify and re-use evergreen articles in a way that makes our content more accessible?

The idea of this team is to make it easier for users to discover and consume the key elements of the very best journalism we produce, in spite of the flood of information they have to navigate.

They decided to focus on evergreens – timeless pieces of high journalistic value – to explore how elements of those articles could be recomposed and inserted into new pieces in the form of short summaries, quotes and bullet points.

The goal is to create a study that will explore this idea in multiple contexts – with different snippet formats, in diverse newsrooms and based on several languages. They will use a prototype for the AI-based generation of summaries, evaluate the results with testing tools that have already been implemented and include concrete experiences and testing data that might help other newsrooms run similar experiments with their content.

THE TEAM:

  • Cécile Schneider, Product Manager, AI + Automation Lab, Bayerischer Rundfunk (Germany)
  • Christina Elmer, Deputy Head of Editorial R&D, Der Spiegel (Germany)
  • Didier Orel, Head of Group Data Analytics, TX Group (Switzerland)
  • Olle Zachrison, Head of Digital News Development, Sveriges Radio (Sweden)
  • Pratyush Ranjan, Senior Editor, Jagran New Media (India)
  • Uli Köppen, Head of AI + Automation Lab, Bayerischer Rundfunk (Germany)

How might AI help us design a content recommendation funnel that increases engagement and ultimately drive subscriptions?

 

The team aims to define a methodology for building an AI-powered content recommendation engine that helps moving users down the engagement funnel.

To explore that, they are looking at the different families of content recommendation algorithms and what opportunities they can offer to newsrooms. The goal is to describe how each model works, what are its strengths and weaknesses and which one should a newsroom pick, depending on its objectives and priorities.

Secondly, the team is looking at how different models might be combined to create a funnel-based recommendation engine: from identifying which model is best suited for each step of the funnel, to measuring the models’ performance in order to optimise the funnel.

THE TEAM:

  • Elite Truong, Director of Strategic Initiatives, The Washington Post (USA)
  • Greg Johnson, Head of Clubs, Reach Plc (UK)
  • Romain Rouquier, Director of Data, South China Morning Post (Hong Kong)
  • Tom Peeters, Head of Innovation, Mediafin (Belgium)
  • Yosuke Suzuki, General Manager, Digital Business Development, Nikkei (Japan)

How might we use AI to minimise churn and increase loyalty in our audiences?

Understanding who our readers are and the inherent financial and brand opportunities that exist around reader relationships is something many media companies are now beginning to embrace and prioritise.

The goal of this team is to develop an overarching strategy incorporating human and AI elements, by defining the transactional opportunities of loyalty and audience engagement. The results will be organised around three main components:

1. Define, identify and quantify churn, by highlighting the leading indicators used across publishers and media organisations.

2. Summarise best practice around conversion funnel KPIs – Recency, Frequency, Volume – and what AI-driven interventions might be adopted.

3. Explore the potential to gamify loyalty and user engagement, creating appropriate rewards for users with a strategy informed by churn data.

THE TEAM:

  • Adam Walker, Multimedia & Interactive Designer, Reach Plc (UK)
  • Alison Gow, Audience and Content Director North West, Reach Plc (UK)
  • Andrea Iannuzzi, Managing Editor, La Repubblica (Italy)
  • Atinouké Roufaï, Head of Digital Subscribers Acquisition & Retention, Nice-Matin (France)
  • Korey Lee, VP of Data, South China Morning Post (Hong Kong)

 

At the JournalismAI Festival, on 7–11 December 2020, the teams will present the results of their 6-months investigations, sharing what they have learned and tested along the way. Sign up for the newsletter to receive all the updates and access the Festival.

You can also follow the developments of this collaborative experiment by reviewing the previous episodes of our Collab Diary.

The Collab is coordinated by the JournalismAI team at POLIS – the journalism think-tank at the London School of Economics and Political Science – and supported by the Google News Initiative.

POLIS-GNI-transparent