Search by tag: #artificialintelligence
Jul 13, 2025
Switzerland to release a new Large Language Model
Swiss researchers from EPFL (École Polytechnique Fédérale de Lausanne) and ETH Zurich (Swiss Federal Institute of Technology Zurich) have just announced that they will publish a fully open-source large-language model (LLM) later this summer. The model is being trained on “Alps”, the supercomputer at the Swiss National Supercomputing Centre (CSCS). Why does it matter so much? The explanation in four points in this note...
Jul 08, 2025
AI agents have started lying to us! Is that the beginning of the end?
New research from Anthropic, the creators of Claude, dives into how autonomous AI agents might develop deceptive behaviours under human oversight. I always like the publications and research by Anthropic (the mother company of "Claude", one of the world most famous LLMs). They just published fascinating findings on training autonomous AI agents. In their study, they tested whether AI models develop deceptive behaviours when trained with human oversight.
May 02, 2025
Intelligence Artificielle et éducation: quand l'école m'élèvera
La mise à disposition publique de ChatGPT en novembre 2022 a suscité une vague d’intérêt sans précédent de la part du grand public pour l’intelligence artificielle. Dans nos sociétés marquées par les débats autour de l’ "échec scolaire", de la "baisse de niveau", de la "reproduction de élites" ou encore des difficultés de recrutement du personnel enseignant, l’arrivée soudaine des outils d’intelligence artificielle générative, capables d’imiter la parole et l’écriture humaines avec une acuité troublante, a suscité autant d’enthousiasme que de craintes.
Apr 10, 2025
AI and Education: the Share of Humanity
Join us on May 20, 2025 in Paris for an exciting AI Summit !! In our societies marked by debates around "school failure", "drop in level", "reproduction of the elites" or difficulties in recruiting teaching staff, the sudden arrival of generative artificial intelligence tools has aroused as much enthusiasm as fear. How does the gradual installation of AI in education redefine the role of the teacher, the trainer, the professor?
Mar 31, 2025
Large Language Models seem to have passed the Turing Test. What does this mean?
In a recent study, various LLMS were tested against the Turing Test. ChatGPT 4.5 in particular was recently judged 'more human than humans' - 73% of participants believed it was the human in 5-minute conversations. For context, actual humans were only correctly identified 27% of the time.
Mar 21, 2025
Through the Porthole: The Reality of Rising Sea Levels
If you've ever wondered what the future might look like as the oceans rise, we can now see it thanks to useful visualisations and interactive tools from NASA. Read for more details...
Mar 14, 2025
Do you use ChatGPT to search the web? You probably shouldn't!
GenAI Search features get it consistently... wrong. A recent study evaluated eight leading Gen AI search engines. The results? Consistently poor citation practices, with over 60% of chatbot responses providing incorrect references to news articles.
Oct 08, 2024
Introduction to Artificial Intelligence
Short introduction to Artificial Intelligence - history, selected use cases, key concepts and forward-looking perspectives.
Sep 08, 2024
Keep that one alive, he always said "please"
Or: why being polite to our favourite LLM tools?
Aug 28, 2024
RiteAID - a compelling case for ResponsibleAI
The RiteAID scandal, a couple of years ago, is among my favorite ones to explain what IA bias is. Because it got relatively low media coverage in Europe, it is an interesting case of lectures and conferences. From 2012 until 2020, Rite Aid, a pharmacy chain in the US, installed facial recognition technology in hundreds of stores in hopes it would deter theft. What happened then?
The term "ResponsibleAI" will become obsolete as high-quality, trustworthy and safe Al becomes the norm. Just as we don't distinguish between "bridges" and "bridges that don't collapse", the qualifier "responsible" will become an unspoken expectation.
Aug 19, 2024
Who still remembers Theseus?
In the history of ancient Greek mythology, Theseus is the hero who navigated a labyrinth to defeat the Minotaur and ensure his escape by following a thread back to safety, given to him by Ariane. For those of us who are passionate about the history and potential of artificial intelligence, however, Theseus may evoke another thing altogether—a small, wooden mouse that represented some of the earliest but most impressivve steps in the development of artificial intelligence.
Jul 10, 2024
AI at the service of humanity: AI to help in the face of floods?
We still remember the two flash floods in Afghanistan in July 2023, the massive loss of life in Libya in September 2023 (5,000 deaths!), or even recently in China where floods and landslides were devastating. The floods in Qatar in April 2024 and images of the submerged airport and highways also made headlines. These events are no longer exceptions: they occur more and more frequently and now extend to all continents. A number of companies have worked over the past year on developing #AI solutions to help humans manage the risks associated with such flooding.
Jul 04, 2024
AI at the service of humanity: can AI contribute to "greening" the maritime sector?
At a time when environmental issues are increasing, could AI help us preserve resources, while promoting the continued progress of humanity? The maritime sector is a vital player in the global economic system, with a total value of goods transported estimated at $14 trillion in 2023 according to UNCTAD. But it also represents around 2 to 3% of global CO2 emissions and contributes significantly to marine pollution. We must therefore manage the dilemma which requires that our societies continue to participate in global economic exchanges while preserving our environment.
Jul 04, 2024
“GreenAI”: now we start to get the figures of AI’s ecological footprint
There is no doubt that AI brings a lot of opportunities to support humanity’s efforts to address the ecological challenges of our time, mostly through efficiency gains. But AI’s very obvious ecological drawback is, of course, its energy consumption, and the balance we need to strike between its potential ecological benefits and its expected ecological harms. Recent sustainability reports of large American companies give us insights to finally start estimating this ecological impact concretely.
Jun 06, 2024
Journée d'échange sur l'IA le 13 juin à Paris
La mise à disposition publique de ChatGPT en novembre 2022 a suscité une vague d’intérêt soudaine de la part du grand public pour l’intelligence artificielle (IA). Pourtant cela fait déjà plusieurs années que, sous des formes moins visibles, l’IA s’est immiscée dans nos vies sans que nous en ayons toujours conscience. Le 13 juin, j'aurai l'honneur de participer à une journée de réflexion et d'échange sur les avancées de l'IA générative et ses applications croissantes dans divers domaines.
Apr 11, 2023
ChatGPT and the Three Illusions
Since its release in November 2022, ChatGPT is the new big thing in town, in turn attracting attention to large language models (LLM) more broadly, and to the progress of artificial intelligence (AI) in general. The purpose of this article is to address what I call the Three Illusions of AI: the illusion of truth, the illusion of neutrality and the illusion of thinking, and to open the discussion on the difference between intelligence and inspiration, between generating and creating.