
Journalist/news editor. MA/MSc Journalism (Digital Journalism) convenor @GoldsmithsUoL. Senior Lecturer. Computational Journalism/MoJo/News+Power @thetimes alum
London
Joined June 2011
- Tweets4.6K
- Following4.2K
- Followers1.7K
- Likes2.9K
Pinned Tweet
We are still accepting applications for our cutting-edge MA/MSc Journalism at Goldsmiths starting this September @GoldsmithsUoL #digitaljournalism #investigativejournalism #datajournalism #aijournalism #mobilejournalism #multimediajournalism gold.ac.uk/pg/ma-journalism-…
Another fantastic CIJ Logan Symposium at the Barbican this year. Congrats to the CIJ team #CIJLogan26
1
273
The Times publisher deploys AI legaltech support lawyers – and reporters news.uk.cityam.com/story/246…
232
“I was sentenced as a terrorist. Now my barrister is on trial for defending me.”opendemocracy.net/email/1236…
248
Helena Bengtsson on using AI in data journalism investigations at #CIJSUMMER 2026. Another fantastic Investigative Journalism Conference by @cijournalism
Goldsmiths MA/MSc Journalism lecturer Jonathan Stoneman talking about using AI for data journalism #CIJSummer at @GoldsmithsUoL gold.ac.uk/pg/ma-journalism-…
74
Great to see Goldsmiths Masters in Digital Journalism alum @m_portocarrero presenting on Multimedia Skills for Your Investigation for #cijsummer at Goldsmiths @MediaComGold @GoldsmithsUoL
1
298
Ted Turner obituary: founder of CNN
thetimes.com/article/d155901…
52
Miranda McLachlan retweeted
Fantastic work by @MediaComGold MA Journalism (Digital Journalism) graduate Patricia Clarke. Come and study journalism with us this Autumn. New modules on critical AI and data this year. gold.ac.uk/pg/ma-journalism/ contact me to find you more (#AI, #criticalAI)
The Observer issue no. 12,228: Bad chat by @paticlarke
How unregulated AI is causing a new mental health crisis
Featuring:
Spring cravings: @NigelSlater and @Tomos_PP recipes in @ObsMagazine
American prayer: Megan Nolan on the Pope v Trump and The US is no longer our friend by Max Hastings
Typhoid Mandy: @CatNeilan on the casualties of the Mandelson appointment and Rachel Sylvester on the civil service turning on Starmer
Plus:
Mind the gap: Arsenal go into City showdown with title hopes on the line
The true confessions of Will Self, as told to Seán O’Hagan in @ObsNewReview
Luella Bartley's art of survival @ObsMagazine
Pick up a copy today, available nationwide. Subscribe to get a copy of The Observer delivered right to your door every week - link in bio.
‘It started with a tip-off’: how a Guardian investigation exposed child sex trafficking on Facebook and Instagram theguardian.com/global-devel…
67
Miranda McLachlan retweeted
Today we published a report on younger news audiences. Authored by Craig T. Robertson, @amyrossarguedas.bsky.social @mitalilive.bsky.social and @richardfletcher.bsky.social, it's based on DNR and other surveys
Full report
reutersinstitute.politics.ox…
Seven findings in thread
Jürgen Habermas obituary: philosopher who was Germany’s moral conscience
thetimes.com/article/08a2187…
1
388
LLMs process text from left to right — each token can only look back at what came before it, never forward. This means that when you write a long prompt with context at the beginning and a question at the end, the model answers the question having "seen" the context, but the context tokens were generated without any awareness of what question was coming. This asymmetry is a basic structural property of how these models work.
The paper asks what happens if you just send the prompt twice in a row, so that every part of the input gets a second pass where it can attend to every other part. The answer is that accuracy goes up across seven different benchmarks and seven different models (from the Gemini, ChatGPT, Claude, and DeepSeek series of LLMs), with no increase in the length of the model's output and no meaningful increase in response time — because processing the input is done in parallel by the hardware anyway.
There are no new losses to compute, no finetuning, no clever prompt engineering beyond the repetition itself.
The gap between this technique and doing nothing is sometimes small, sometimes large (one model went from 21% to 97% on a task involving finding a name in a list). If you are thinking about how to get better results from these models without paying for longer outputs or slower responses, that's a fairly concrete and low-effort finding.
Read with AI tutor: chapterpal.com/s/1b15378b/pr…
Get the PDF: arxiv.org/pdf/2512.14982
Miranda McLachlan retweeted
How will AI reshape the future of news in 2026?
We asked 17 media managers working in this field, including @rubinafillion @OlleZ @ezraeeman @sannuta @jaemark and others
Click to read
reutersinstitute.politics.ox…
Miranda McLachlan retweeted
ICYMI: The Wayfinder 2025-26 Review/Preview Report is out, connecting dots across media, technology, culture, and power.
Across the chapters: browsing gives way to asking, content turns fluid, attention fragments yet platforms consolidate…
linkedin.com/posts/ezra-eema…
We are accepting applications for our cutting-edge Masters in Digital Journalism at Goldsmiths starting this September @GoldsmithsUoL @GoldsmithsMADJ gold.ac.uk/pg/ma-digital-jou…
125
Miranda McLachlan retweeted
Join us this evening at @mediacomgold @GoldsmithsUoL for a fascinating discussion and film-screening on how to find the tools to do the essential work of #journalism. Details 👇🏽 complete with empanadas!
There’s a livestream too on this link: youtube.com/live/IHf4IueVvh8
15 essential tasks GPTs can do for journalists journalism.co.uk/news/how-ca…
53
Journalists among 20 killed in Israeli strike on Gaza hospital
thetimes.com/article/e2f7fd0…
65




















