Luc Rocher
@rocher.lc
associate professor at Oxford · UKRI future leaders fellow · i study how data and algorithms shape societies · AI fairness, accountability and transparency · algorithm auditing · photographer, keen 🚴🏻 · they/them · https://rocher.lc (views my own)
Reposted by Luc Rocher
ICYMI: Great piece by @decorrespondent.nl including an interview with @rocher.lc about AI benchmarks, the way main AI models are currently evaluation and their latest co-authored review of 445 AI benchmarks. Read the full article: decorrespondent.nl/16539/de-cla...
De claims over steeds slimmere AI-modellen? Meer vibe dan wetenschap
Benchmarks, de heilige rapportcijfers van AI-modellen, worden opgevoerd als wetenschappelijk bewijs om aan te tonen dat AI steeds intelligenter wordt. Maar is dat wel zo? Nieuw onderzoek laat zien dat...
decorrespondent.nl
November 19, 2025 at 12:34 PM
ICYMI: Great piece by @decorrespondent.nl including an interview with @rocher.lc about AI benchmarks, the way main AI models are currently evaluation and their latest co-authored review of 445 AI benchmarks. Read the full article: decorrespondent.nl/16539/de-cla...
Reposted by Luc Rocher
A bunch of people thinking that AI will ‘help’ us add alt-text to graphs and images to improve accessibility better come up with a real plan instead (if we shouldn’t have already known that already for one reason or another).
📸 New research just out revealing disturbing problems with AI image captioning. AI models can now analyse both texts and images, and are trained on vast collections of human-created content spanning centuries. Some claims these models could soon help human historians interpret and explain the past.
November 17, 2025 at 4:32 PM
A bunch of people thinking that AI will ‘help’ us add alt-text to graphs and images to improve accessibility better come up with a real plan instead (if we shouldn’t have already known that already for one reason or another).
📸 New research just out revealing disturbing problems with AI image captioning. AI models can now analyse both texts and images, and are trained on vast collections of human-created content spanning centuries. Some claims these models could soon help human historians interpret and explain the past.
November 15, 2025 at 12:13 PM
📸 New research just out revealing disturbing problems with AI image captioning. AI models can now analyse both texts and images, and are trained on vast collections of human-created content spanning centuries. Some claims these models could soon help human historians interpret and explain the past.
Reposted by Luc Rocher
Scientists and scholars in AI and its social impacts call on von der Leyen to retract #AIHype statement.
@olivia.science
@abeba.bsky.social
@irisvanrooij.bsky.social
@alexhanna.bsky.social
@rocher.lc
@danmcquillan.bsky.social
@robin.berjon.com
& many others have signed
www.iccl.ie/press-releas...
@olivia.science
@abeba.bsky.social
@irisvanrooij.bsky.social
@alexhanna.bsky.social
@rocher.lc
@danmcquillan.bsky.social
@robin.berjon.com
& many others have signed
www.iccl.ie/press-releas...
Scientists call on the President of the European Commission to retract AI hype statement
Experts in AI call on the President of the European Commission to retract unscientific AI hype statement she made in the budget speech.
www.iccl.ie
November 10, 2025 at 9:48 AM
Scientists and scholars in AI and its social impacts call on von der Leyen to retract #AIHype statement.
@olivia.science
@abeba.bsky.social
@irisvanrooij.bsky.social
@alexhanna.bsky.social
@rocher.lc
@danmcquillan.bsky.social
@robin.berjon.com
& many others have signed
www.iccl.ie/press-releas...
@olivia.science
@abeba.bsky.social
@irisvanrooij.bsky.social
@alexhanna.bsky.social
@rocher.lc
@danmcquillan.bsky.social
@robin.berjon.com
& many others have signed
www.iccl.ie/press-releas...
Reposted by Luc Rocher
Please sign this excellent letter to Ursula vdL, so she can see that nonsense AI hype is neither true nor appropriate for anybody to believe or spread.
Scientists and scholars in AI and its social impacts call on von der Leyen to retract #AIHype statement.
@olivia.science
@abeba.bsky.social
@irisvanrooij.bsky.social
@alexhanna.bsky.social
@rocher.lc
@danmcquillan.bsky.social
@robin.berjon.com
& many others have signed
www.iccl.ie/press-releas...
@olivia.science
@abeba.bsky.social
@irisvanrooij.bsky.social
@alexhanna.bsky.social
@rocher.lc
@danmcquillan.bsky.social
@robin.berjon.com
& many others have signed
www.iccl.ie/press-releas...
Scientists call on the President of the European Commission to retract AI hype statement
Experts in AI call on the President of the European Commission to retract unscientific AI hype statement she made in the budget speech.
www.iccl.ie
November 10, 2025 at 12:16 PM
Please sign this excellent letter to Ursula vdL, so she can see that nonsense AI hype is neither true nor appropriate for anybody to believe or spread.
Reposted by Luc Rocher
Stop overhyping AI, scientists tell von der Leyen
@euractiv.com @maximilian-henning.bsky.social
www.euractiv.com/news/stop-ov...
@euractiv.com @maximilian-henning.bsky.social
www.euractiv.com/news/stop-ov...
Stop overhyping AI, scientists tell von der Leyen | Euractiv
Open letter criticises the Commission President for remarks earlier this year – when she anticipated AI would “approach human reasoning” in 2026
www.euractiv.com
November 10, 2025 at 10:12 AM
Stop overhyping AI, scientists tell von der Leyen
@euractiv.com @maximilian-henning.bsky.social
www.euractiv.com/news/stop-ov...
@euractiv.com @maximilian-henning.bsky.social
www.euractiv.com/news/stop-ov...
Is there a scientific crisis in AI evaluations‽ We did the hard work of reviewing all recent AI benchmarks in top AI conferences. Amongst the 445 reviewed benchmarks, few offer rigorous evaluations of AI capabilities, making it difficult to actually track advances in the field and compare models.
November 4, 2025 at 12:03 PM
Is there a scientific crisis in AI evaluations‽ We did the hard work of reviewing all recent AI benchmarks in top AI conferences. Amongst the 445 reviewed benchmarks, few offer rigorous evaluations of AI capabilities, making it difficult to actually track advances in the field and compare models.
Reposted by Luc Rocher
In short, the problem here is not that Signal ‘chose’ to run on AWS. The problem is the concentration of power in the infrastructure space that means there isn’t really another choice: the entire stack, practically speaking, is owned by 3-4 players. 11/
October 27, 2025 at 10:38 AM
In short, the problem here is not that Signal ‘chose’ to run on AWS. The problem is the concentration of power in the infrastructure space that means there isn’t really another choice: the entire stack, practically speaking, is owned by 3-4 players. 11/
Reposted by Luc Rocher
Our paper is out! 🎉
As part of my research at @oii.ox.ac.uk under the supervision of @rocher.lc and @computermacgyver.bsky.social, I built a language model to help human rights analysts crowdsource reports of gun violence.
The paper was recently published at ACM CSCW 👇
dl.acm.org/doi/10.1145/...
As part of my research at @oii.ox.ac.uk under the supervision of @rocher.lc and @computermacgyver.bsky.social, I built a language model to help human rights analysts crowdsource reports of gun violence.
The paper was recently published at ACM CSCW 👇
dl.acm.org/doi/10.1145/...
Into the Crossfire: Evaluating the Use of a Language Model to Crowdsource Gun Violence Reports | Proceedings of the ACM on Human-Computer Interaction
Gun violence is a pressing human rights issue that affects nearly every dimension
of the social fabric, from healthcare and education to psychology and the economy.
Reliable data on firearm events is ...
dl.acm.org
October 27, 2025 at 9:17 PM
Our paper is out! 🎉
As part of my research at @oii.ox.ac.uk under the supervision of @rocher.lc and @computermacgyver.bsky.social, I built a language model to help human rights analysts crowdsource reports of gun violence.
The paper was recently published at ACM CSCW 👇
dl.acm.org/doi/10.1145/...
As part of my research at @oii.ox.ac.uk under the supervision of @rocher.lc and @computermacgyver.bsky.social, I built a language model to help human rights analysts crowdsource reports of gun violence.
The paper was recently published at ACM CSCW 👇
dl.acm.org/doi/10.1145/...
Reposted by Luc Rocher
After submitting a FOIA request UKRI, I obtained success rates by three grant call scheme and I can only say that I am disheartened by the results:
- AHRC Responsive Mode 2025: 2%
- ESRC New Investigator Grant 2025: 1%
- ESRC Research Grant Round 2025: 1%
- AHRC Responsive Mode 2025: 2%
- ESRC New Investigator Grant 2025: 1%
- ESRC Research Grant Round 2025: 1%
October 22, 2025 at 7:55 AM
After submitting a FOIA request UKRI, I obtained success rates by three grant call scheme and I can only say that I am disheartened by the results:
- AHRC Responsive Mode 2025: 2%
- ESRC New Investigator Grant 2025: 1%
- ESRC Research Grant Round 2025: 1%
- AHRC Responsive Mode 2025: 2%
- ESRC New Investigator Grant 2025: 1%
- ESRC Research Grant Round 2025: 1%
Reposted by Luc Rocher
This morning, I’m at the CSCW conference in a session on “Activism in a Time if Data.” Most of the papers seem to assume activism for “good,” ending with design recs for supporting activism. As far as I can tell, none of the papers center right wing activism. Seems like we have a massive blind spot.
October 21, 2025 at 9:25 AM
This morning, I’m at the CSCW conference in a session on “Activism in a Time if Data.” Most of the papers seem to assume activism for “good,” ending with design recs for supporting activism. As far as I can tell, none of the papers center right wing activism. Seems like we have a massive blind spot.
Reposted by Luc Rocher
AI usage accounted for less than 1% of online activity and “the most consistent predictors of AI use across studies were…Machiavellianism, narcissism, psychopathy,” i.e., dark triad personality traits. Nice. www.liebertpub.com/doi/10.1177/...
Evaluating Artificial Intelligence Use and Its Psychological Correlates via Months of Web-Browsing Data | Cyberpsychology, Behavior, and Social Networking
Despite widespread discussions about artificial intelligence (AI) and its impact on society, little work has objectively measured how often people use this technology in the wild. The present article ...
www.liebertpub.com
October 12, 2025 at 5:43 PM
AI usage accounted for less than 1% of online activity and “the most consistent predictors of AI use across studies were…Machiavellianism, narcissism, psychopathy,” i.e., dark triad personality traits. Nice. www.liebertpub.com/doi/10.1177/...
Reposted by Luc Rocher
Dr. Jane Goodall filmed an interview with Netflix in March 2025 that she understood would only be released after her death.
October 5, 2025 at 9:08 AM
Dr. Jane Goodall filmed an interview with Netflix in March 2025 that she understood would only be released after her death.
Reposted by Luc Rocher
« Ils ont traîné la petite Greta [Thunberg] par les cheveux sous nos yeux, l'ont battue et l'ont forcée à embrasser le drapeau israélien. Ils lui ont fait tout ce qu'ils pouvaient imaginer, en guise d'avertissement pour les autres » www.theguardian.com/world/2025/o...
Israel accused of detaining Greta Thunberg in infested cell and making her hold flags
Activist tells Swedish officials she has been subjected to harsh treatment, including insufficient food and water
www.theguardian.com
October 4, 2025 at 9:18 PM
« Ils ont traîné la petite Greta [Thunberg] par les cheveux sous nos yeux, l'ont battue et l'ont forcée à embrasser le drapeau israélien. Ils lui ont fait tout ce qu'ils pouvaient imaginer, en guise d'avertissement pour les autres » www.theguardian.com/world/2025/o...
Reposted by Luc Rocher
The Saudi government murdered Jamal Khashoggi seven years ago today. His editor @karenattiah.bsky.social wrote this two years ago.
Opinion | The moment I knew the world would try to move on from Jamal Khashoggi
I could see it coming at Time’s 2018 Person of the Year celebration.
wapo.st
October 2, 2025 at 12:06 PM
The Saudi government murdered Jamal Khashoggi seven years ago today. His editor @karenattiah.bsky.social wrote this two years ago.
Reposted by Luc Rocher
Perhaps the final meeting from the old @oii.ox.ac.uk buildings before the institute moves to its new haunt at the Schwarzman Centre for the Humanities – featuring the great Chico Kerche, @rocher.lc, Franziska Hafner and @lujain.bsky.social to talk about information and news within LLMs.
September 5, 2025 at 3:16 PM
Perhaps the final meeting from the old @oii.ox.ac.uk buildings before the institute moves to its new haunt at the Schwarzman Centre for the Humanities – featuring the great Chico Kerche, @rocher.lc, Franziska Hafner and @lujain.bsky.social to talk about information and news within LLMs.
Reposted by Luc Rocher
Incredible scenes as the University of Warwick refuses to answer an FOI on the cost of an advertorial photo feature on their chief marketing officer in Vogue Singapore because it is a 'trade secret' www.whatdotheyknow.com/request/cost...
Cost of Vogue Singapore Advertorial - a Freedom of Information request to University of Warwick
I would like to request 1) all email correspondence relating to, and 2) the price to the University of Warwick of, the placement of the following article in Vogue Singapore:
Ajay Teli on how cultural...
www.whatdotheyknow.com
August 24, 2025 at 11:35 AM
Incredible scenes as the University of Warwick refuses to answer an FOI on the cost of an advertorial photo feature on their chief marketing officer in Vogue Singapore because it is a 'trade secret' www.whatdotheyknow.com/request/cost...
Reposted by Luc Rocher
Facial recognition systems often boast near-perfect test scores, but real life tells a different story. OII researchers Teo Canmetin, Juliette Zaccour @biksil.bsky.social and Luc Rocher @rocher.lc expose the gap between lab evaluations and real-world conditions.
Lab evaluations of facial recognition claiming a high level of accuracy have been used to justify the technology's vast deployment, but they often ignore how the software could perform in diverse, messy and unpredictable real-world environments, Teo Canmetin, Juliette Zaccour and Luc Rocher write.
Why We Shouldn’t Trust Facial Recognition’s Glowing Test Scores | TechPolicy.Press
While lab evaluations of facial recognition may appear objective, they often ignore real-world performance, write Teo Canmentin, Juliette Zaccour & Luc Rocher.
www.techpolicy.press
August 19, 2025 at 9:36 AM
Facial recognition systems often boast near-perfect test scores, but real life tells a different story. OII researchers Teo Canmetin, Juliette Zaccour @biksil.bsky.social and Luc Rocher @rocher.lc expose the gap between lab evaluations and real-world conditions.
Reposted by Luc Rocher
Research by @rocher.lc @lujain.bsky.social and Sofia Hafner cited in the Telegraph shows the warmer a chatbot feels, the more it may drift into conspiracy. OpenAI’s fix shows how hard it is to balance empathy & truth in AI. www.telegraph.co.uk/business/202...
August 19, 2025 at 8:52 AM
Research by @rocher.lc @lujain.bsky.social and Sofia Hafner cited in the Telegraph shows the warmer a chatbot feels, the more it may drift into conspiracy. OpenAI’s fix shows how hard it is to balance empathy & truth in AI. www.telegraph.co.uk/business/202...
Reposted by Luc Rocher
Lab evaluations of facial recognition claiming a high level of accuracy have been used to justify the technology's vast deployment, but they often ignore how the software could perform in diverse, messy and unpredictable real-world environments, Teo Canmetin, Juliette Zaccour and Luc Rocher write.
Why We Shouldn’t Trust Facial Recognition’s Glowing Test Scores | TechPolicy.Press
While lab evaluations of facial recognition may appear objective, they often ignore real-world performance, write Teo Canmentin, Juliette Zaccour & Luc Rocher.
www.techpolicy.press
August 18, 2025 at 5:56 PM
Lab evaluations of facial recognition claiming a high level of accuracy have been used to justify the technology's vast deployment, but they often ignore how the software could perform in diverse, messy and unpredictable real-world environments, Teo Canmetin, Juliette Zaccour and Luc Rocher write.
Facial recognition just passed 99.95% accuracy in the latest evaluations by the US National Institute of Standards and Technology.
But did you know that these numbers come from pristine lab conditions with small datasets, good lightning, and clear photos?
But did you know that these numbers come from pristine lab conditions with small datasets, good lightning, and clear photos?
August 18, 2025 at 4:19 PM
Facial recognition just passed 99.95% accuracy in the latest evaluations by the US National Institute of Standards and Technology.
But did you know that these numbers come from pristine lab conditions with small datasets, good lightning, and clear photos?
But did you know that these numbers come from pristine lab conditions with small datasets, good lightning, and clear photos?
Reposted by Luc Rocher
fascinating paper on 'character training' LLMs - models fine-tuned to be 'warmer' are significantly more likely to promote conspiracy theories or validate incorrect facts, particularly when users express sadness 🤯 arxiv.org/abs/2507.21919
Training language models to be warm and empathetic makes them less reliable and more sycophantic
Artificial intelligence (AI) developers are increasingly building language models with warm and empathetic personas that millions of people now use for advice, therapy, and companionship. Here, we sho...
arxiv.org
July 31, 2025 at 3:29 PM
fascinating paper on 'character training' LLMs - models fine-tuned to be 'warmer' are significantly more likely to promote conspiracy theories or validate incorrect facts, particularly when users express sadness 🤯 arxiv.org/abs/2507.21919
🧵New research! People now use AI chatbots for therapy, friendship, and romance. These days, both specialised and general-purpose chatbots are built with a friendly, empathetic tone to best comfort users. We find this new trend is not just cosmetic but seriously harms user safety.
August 1, 2025 at 11:51 AM
🧵New research! People now use AI chatbots for therapy, friendship, and romance. These days, both specialised and general-purpose chatbots are built with a friendly, empathetic tone to best comfort users. We find this new trend is not just cosmetic but seriously harms user safety.
Reposted by Luc Rocher
Our new investigation: Ireland's own goal
Almost every department of the Irish Government and local authorities have installed Chinese surveillance systems.
It is hard to travel in Ireland without being tracked by China’s Hikvision surveillance system.
www.iccl.ie/digital-data...
Almost every department of the Irish Government and local authorities have installed Chinese surveillance systems.
It is hard to travel in Ireland without being tracked by China’s Hikvision surveillance system.
www.iccl.ie/digital-data...
Ireland's own goal
ICCL Enforce investigation finds Hikvision surveillance equipment procured and installed across Irish bodies. The Irish State pays for China's surveillance in Ireland.
www.iccl.ie
July 23, 2025 at 7:36 AM
Our new investigation: Ireland's own goal
Almost every department of the Irish Government and local authorities have installed Chinese surveillance systems.
It is hard to travel in Ireland without being tracked by China’s Hikvision surveillance system.
www.iccl.ie/digital-data...
Almost every department of the Irish Government and local authorities have installed Chinese surveillance systems.
It is hard to travel in Ireland without being tracked by China’s Hikvision surveillance system.
www.iccl.ie/digital-data...
We keep seeing these articles suggesting privacy is dead and we can all be tracked by all sort of apparatuses. We did the maths earlier this year in an article in Nature Communications, and none of this holds.
Humans can be tracked with unique 'fingerprint' based on how their bodies block Wi-Fi signals - www.theregister.com/2025/07/22/w... "Observers could therefore track a person as they pass through signals sent by different Wi-Fi networks – even if they’re not carrying a phone." #surveillance #privacy
WhoFi: Unique 'fingerprint' based on Wi-Fi interactions
: Wi-Fi spy with my little eye that same guy I saw at another hotspot
www.theregister.com
July 23, 2025 at 2:50 AM
We keep seeing these articles suggesting privacy is dead and we can all be tracked by all sort of apparatuses. We did the maths earlier this year in an article in Nature Communications, and none of this holds.