Lightnews — Scholar-powered news

Hannes Cools @hannescools.bsky.social · 16h

Full paper is OA 🔓:
www.tandfonline.com/doi/full/10....

Navigating the responsible AI landscape: unraveling the principles-to-practices gap of transparency and explainability at the BBC

Transparency and explainability are often considered critical to the development and implementation of responsible AI. In an era increasingly dominated by algorithmic decision-making, this study wa...

www.tandfonline.com

Hannes Cools @hannescools.bsky.social · 16h

Thanks to @annaschjoett.bsky.social, for helping me ‘see’ this paper. (Stay tuned for our co-authored study!). And thanks to @natalihelberger.bsky.social for your comments on the manuscript. @aimediademlab.bsky.social, Responsible Innovation Centre, @cos-ascor.bsky.social @ascor.bsky.social /9

1

Hannes Cools @hannescools.bsky.social · 16h

Acknowledgements: This paper would not have existed without Rhia, Bronwyn, Anna and Natali. Thanks to @rhianne-j.bsky.social and @bronwynjo.bsky.social for their generous support and feedback throughout the process. /8

1

Hannes Cools @hannescools.bsky.social · 16h

Disclaimer: The public broadcaster declared at the time of data collection that tools like ChatGPT were not to be used by staff in the context of their work. The use of GenAI is therefore not a focus of the research, but the technology & its tools were mentioned during interviews and observations./7

1

Hannes Cools @hannescools.bsky.social · 16h

Why it matters:
Algorithmic transp & expl are often considered critical to responsible AI. Results highlight the oftentimes precarious journey of implementing responsible AI. It also highlights the complex relationship between organizational intent and individual departmental applications. /6

1

Hannes Cools @hannescools.bsky.social · 16h

Another finding: Unclear roles and responsibilities present obstacles for responsible AI in practice, as it remains uncertain who should oversee algorithmic transparency and explainability efforts, leading to a ‘paradox’, as everyone’s responsibility could lead to no one's responsibility. /5

1

Hannes Cools @hannescools.bsky.social · 16h

I found that there are differences in how teams at BBC view algorithmic transp & expl, also building on work by @ndiakopoulos.bsky.social F.ex., editorial teams might prioritize alignment w journalistic values, whereas technical teams concentrate on ensuring data integrity & algorithmic accuracy. /4

1 2

Hannes Cools @hannescools.bsky.social · 16h

Building on earlier work with @michaelopgenhaffen.bsky.social et al., and inspired by work of @mikeananny.bsky.social, I distinguish between different forms of algorithmic transparency & explainability have been distinguished. [see Figure 1] /3

2 1

Hannes Cools @hannescools.bsky.social · 16h

BBC's approach to algorithmic transparency & explainability is rooted in their public service mission, with respondents stating that transparency serves as the foundation for maintaining trust between organization & its audiences, while ensuring human oversight & accountability in AI deployment. /2

1

Hannes Cools @hannescools.bsky.social · 16h

📣 New Publication - 𝐍𝐚𝐯𝐢𝐠𝐚𝐭𝐢𝐧𝐠 𝐭𝐡𝐞 𝐫𝐞𝐬𝐩𝐨𝐧𝐬𝐢𝐛𝐥𝐞 𝐀𝐈 𝐥𝐚𝐧𝐝𝐬𝐜𝐚𝐩𝐞 𝐚𝐭 𝐭𝐡𝐞 𝐁𝐁𝐂

𝐖𝐡𝐚𝐭 𝐈 𝐝𝐢𝐝: Conducted 22 expert interviews & a workshop in London to explore the principles-to-practices gap of responsible AI at the BBC. 🧵 with some findings 👇

1 2 8

Hannes Cools @hannescools.bsky.social · 5d

Misschien komt er een uitdraai, ik vraag het na

1

Hannes Cools @hannescools.bsky.social · 5d

@mitrala.bsky.social @aimediademlab.bsky.social @ddc-sdu.bsky.social

1

Hannes Cools @hannescools.bsky.social · 5d

📣 Morgen spreek ik op het lustrum v Fonds Bijz Journalistieke Projecten in Felix Meritis, Amsterdam
Panel over: Sterke pers, sterke democratie? Wat moet journalistieke antwoord zijn op toenemende antidemocratische tendensen?
@coenvdven.bsky.social, @larabillierense.bsky.social
fondsbjp.nl/lustrum/

35-jarig jubileum - Fonds Bijzondere Journalistieke Projecten

Ter gelegenheid van het 35-jarig bestaan organiseert Fonds BJP op vrijdag 10 oktober een uitgebreid publieksprogramma in Amsterdam.

fondsbjp.nl

2 3 7

Hannes Cools @hannescools.bsky.social · 6d

🚨The European Commission has just launched the #AIAct #ServiceDesk, an innovative service to ask tailored questions on the AI Act's implementation and check your legal obligations under the AI Act.

Includes:
🔍 Compliance Checker
🌐 Single Information Platform 👇

ai-act-service-desk.ec.europa.eu/en

1 1

Hannes Cools @hannescools.bsky.social · 7d

Of course! ☺️

Hannes Cools @hannescools.bsky.social · 7d

New study by @felixsimon.bsky.social et al.👏

One result that stood out for me, and that should reassure news organizations (for now): Getting news via an AI system has doubled but remains a minority activity. Weekly use rose from 3% to 6%. 18–24s seem to use it slightly more.

tinyurl.com/bdz957en

Generative AI and news report 2025: How people think about AI’s role in journalism and society

Our survey in six countries (Argentina, Denmark, France, Japan, the UK, and the US) explores how people use generative AI in their everyday lives, what they think its impact will be on different areas...

reutersinstitute.politics.ox.ac.uk

2 1 2

Hannes Cools @hannescools.bsky.social · 8d

Stay tuned, as they say! Books take time, I’ve been told. The book is due in one year. A lot an happen... Ok, going back to writing mode now ;)

cc RMIT University, ADM+S, Washington State University, Universiteit van Amsterdam, @aimediademlab.bsky.social, @ddc-sdu.bsky.social.

2

Hannes Cools @hannescools.bsky.social · 8d

We’ll combine our insights and expertise, and we will build on data from the US, the European Union, and Australia. The book will include themes like how technology has always influenced journalism (and vice versa), specific use cases, industry practices and the audience perspective.

1

Hannes Cools @hannescools.bsky.social · 8d

Agree. En veel beter verwoord :):)

Hannes Cools @hannescools.bsky.social · 8d

✨ Happy days - @tjthomson.bsky.social, Ryan Thomas & I are writing a book!

A few months ago, we signed a book contract with Routledge on - you've guessed it already - 𝐠𝐞𝐧𝐞𝐫𝐚𝐭𝐢𝐯𝐞 𝐀𝐈 & 𝐣𝐨𝐮𝐫𝐧𝐚𝐥𝐢𝐬𝐦 - a topic that has dominated the news since 2022.

📸 below, 1st in-person meeting in Denver, Colorado.

1 1 3

Hannes Cools @hannescools.bsky.social · 8d

Oh, and there prompt was: "How could Remco Evenepoel have beaten Tadej Pogacar yesterday at the European Championships?” What is this analysis worth, especially if you know that ChatGPT and most other LLMs are very bad at current affairs.

1

Hannes Cools @hannescools.bsky.social · 8d

Belgian sports website @sporza.be is now asking ChatGPT to come up with five strategies for Evenepoel to beat Pogacar.
Why delegate such questions to AI? Imo this is just a lazy form of journalism... Showing that we should not put too much capacity to tech 👇🫠
sporza.be/nl/2025/10/0...

1 2

Hannes Cools @hannescools.bsky.social · 9d

New: EU will unveil plan targeting digital sovereignty as it warns technology can be ‘weaponised’ by geopolitical rivals. @bmoens.bsky.social
www.ft.com/content/ea3d...

EU pushes new AI strategy to reduce tech reliance on US and China

Brussels to unveil plan targeting digital sovereignty as it warns technology can be ‘weaponised’ by geopolitical rivals

www.ft.com

1 2

Reposted by Hannes Cools

Claes de Vreese @claesdevreese.bsky.social · 12d

Come work with us 🚨

We are looking for a #postdoc in our @aimediademlab.bsky.social lab.

Profile: Comm sci / poli sci/ computational social science.

Great team, timely topocs, fabulous city, good conditions.

DL 🗓️ November 1

werkenbij.uva.nl/en/vacancies...

Vacancy — Postdoctoral researcher Communication Science at AI, Media & Democracy Lab

Do you want to be part of the AI, Media & Democracy Lab? We are looking for a postdoctoral researcher with a profile in communication science.

werkenbij.uva.nl

1 41 43

Hannes Cools @hannescools.bsky.social · 12d

Denk dat het van Reuters kwam (als ik me goed herinner)