Dr. Nicolas Kayser-Bril

What artificial butter tells us about Artificial Intelligence

Platform work regulation is failing cleaners

Researchers studying algorithmic management have long had a blind spot when it comes to cleaning apps’ workers. The evidence now emerging paints a nuanced picture: algorithms play a far smaller role than they do for drivers and couriers, yet cleaners still face severe consequences.

Are all LLMs the same? I set out to measure their similarity.

AI-generated text often has a certain je-ne-sais-quoi of monotony and blandness, whatever model is used. An attempt to put a number on this phenomenon shows once more that measuring meaning is close to impossible.

Why left-wing parties are not pushing for GenAI

I'm often asked why there are so few positive visions of AI. After all, machines can make work less tedious and, provided the benefits are shared fairly, increase the well-being of all. But generative AI does not quite work that way.

An AI slop farm stole my identity

AI slop is everywhere. By one estimate, roughly one in five articles referenced on Google Discover is automatically generated. It's less known that AI slop farms steal the identity of existing journalists and influencers to increase the odds of being picked up by these algorithms.

Remote biometric identification is a colonial boomerang

While many feel – rightly – outraged by how live face recognition and other remote biometric identification infringes on our freedom in public spaces, I think it misses – in part – the point of the technology. To understand its purpose, we need to look into colonial history.

AI avatars have been in politics for a decade, but Diella is different

Albania’s new “Minister for AI” is an LLM-powered avatar. AI has been touted since at least 2017 to replace politicians, but the Albanian software, which shall be in charge of public procurement, opens the gates to corruption on a new scale.

For politicians, TikTok’s algorithm is a handy bogeyman

A parliamentary report about TikTok has received lots of media attention in France in the last few weeks. It is interesting in its own right, and it's also a prime example of how many European politicians behave when faced with an issue related to algorithms.

AI Jesus: What’s left of Pope Francis’ guidance?

A church in Austria put an AI-generated exhibition on display. This does not exactly fit with the Vatican's guidelines on AI.

Image Generators Are Trying to Hide Their Biases – And They Make Them Worse

In the run-up to the EU elections, AlgorithmWatch has investigated which election-related images can be generated by popular AI systems. Two of the largest providers don’t adhere to security measures they have announced themselves recently.

Fighting in the Dark: How Europeans Push Back Against Rogue AI

Automated systems go astray, they contribute to child sexual violence, deny people their social benefits, or block organizations' online presence. The affected people often feel helpless when their rights are violated, but some are taking up the fight while current laws fail to protect the victims.

How Not to: We Failed at Analyzing Public Discourse on AI With ChatGPT

We wanted to collect data on how newspapers wrote about AI. Swayed by the sirens of OpenAI, we thought that using a Large Language Model was the best way to do such a massive text analysis. Alas, we failed. But LLMs are probably not to blame.

The year we waited for action: 2023 in review

Exactly one year ago, I wrote that automated systems might be regulated for good in 2023. This was too optimistic. Not only did European institutions fail to pass the AI Act, even in its watered-down version; the rise of generative models brought us to a new level of danger.

Some image generators produce more problematic stereotypes than others, but all fail at diversity

Automated image generators are often accused of spreading harmful stereotypes, but studies usually only look at MidJourney. Other tools make serious efforts to increase diversity in their output, but effective remedies remain elusive.

AlgorithmWatch welcomes first fellows in algorithmic accountability reporting

After the announcement that we offer a fellowship in algorithmic accountability reporting, we received over 100 applications and were overwhelmed by the quality and diversity of them. After a careful examination, we’re very happy to introduce our first six fellows, an outstanding group of journalists, academics, and a civil society activist.

The year automated systems might have been regulated: 2022 in review

Automated systems were surprisingly absent from this year’s major stories. On the regulation front, European institutions stepped up their efforts. How much change Europeans can expect depends on the institutions’ resolve, and the first test of 2023 already began.

Wolt: Couriers’ feelings don’t always match the transparency report

In August, the Finnish delivery service Wolt published its first “algorithmic transparency report”. We asked three couriers about their experiences. They don't always match the report’s contents.

Algorithmic elections: How automated systems quietly disenfranchise voters

In the United States, automated systems purge voters rolls and verify signatures. They sometimes discard valid votes, especially in historically marginalized communities. Such systems are unlikely to be as widespread in the European Union.

Mastodon could make the public sphere less toxic, but not for all

The open-source social network gained millions of new users following Twitter’s takeover. While some of its features could improve the quality of public discourse, disadvantaged communities might be excluded.

The fediverse is growing, but power imbalances might stay

Alternative social network Mastodon, which has no algorithmic timeline and a decentralized structure, is rapidly gaining steam. But the regulatory framework, which was built for billion-dollar companies, could dampen its growth.

How researchers are upping their game to audit recommender systems

Recommender systems, such as a social network’s news feed or a streaming service’s recommendations, are notoriously difficult to audit. Despite high hurdles, practitioners from journalism and academia are pushing forward.

AlgorithmWatch is offering 5 fellowships in algorithmic accountability reporting

Fellows will receive €1,200 per month during 6 months and will report on automated decision-making in Europe.

Meta is sued for abetting fraud, and they don’t want you to know about it

The Australian consumer commission accuses Meta of helping scammers defraud its users. The scams are global and the legal case could have repercussions elsewhere.

Face recognition data set of trans people still available online years after it was supposedly taken down

A US academic scraped videos off YouTube to train face recognition software on pictures of trans people. New research reveals that his methods were even more carefree than previously known.

A European newsroom to investigate automated systems

AlgorithmWatch is delighted to announce that the Deutsche Postcode Lotterie is funding a decentralized newsroom to investigate automated systems in Europe.