Anastasia Stasenko is the Co-Founder of Pleias, a French startup specializing in developing open-source, copyright-free large language models (LLMs). She holds a PhD in Philosophy and is a Senior Associate Lecturer in Data Analysis and Digital Strategy at Sorbonne-Nouvelle University. Stasenko has led groundbreaking projects, including the creation of CommonCorpus, the largest open dataset for training LLMs, and the development of Albert, an AI assistant for French civil servants. Her work focuses on ethical, transparent, and sovereign AI, with applications in education, healthcare, and public administration. Stasenko is a prominent advocate for open science and frugal AI, demonstrating that high-performance models can be built without copyrighted data. We really hope she's coming back in 2025.
At the last AI Summit in Paris, Pleais announced a new global program to explore https://www.currentai.org
Anastasia Stasenko is the Co-Founder of Pleias, a French startup specializing in developing open-source, copyright-free large language models (LLMs). She holds a PhD in Philosophy and is a Senior Associate Lecturer in Data Analysis and Digital Strategy at Sorbonne-Nouvelle University. Stasenko has led groundbreaking projects, including the creation of CommonCorpus, the largest open dataset for training LLMs, and the development of Albert, an AI assistant for French civil servants. Her work focuses on ethical, transparent, and sovereign AI, with applications in education, healthcare, and public administration. Stasenko is a prominent advocate for open science and frugal AI, demonstrating that high-performance models can be built without copyrighted data. We really hope she's coming back in 2025.
At the last AI Summit in Paris, Pleais announced a new global program to explore https://www.currentai.org
Anastasia Stasenko is the Co-Founder of Pleias, a French startup specializing in developing open-source, copyright-free large language models (LLMs). She holds a PhD in Philosophy and is a Senior Associate Lecturer in Data Analysis and Digital Strategy at Sorbonne-Nouvelle University. Stasenko has led groundbreaking projects, including the creation of CommonCorpus, the largest open dataset for training LLMs, and the development of Albert, an AI assistant for French civil servants. Her work focuses on ethical, transparent, and sovereign AI, with applications in education, healthcare, and public administration. Stasenko is a prominent advocate for open science and frugal AI, demonstrating that high-performance models can be built without copyrighted data. We really hope she's coming back in 2025.
At the last AI Summit in Paris, Pleais announced a new global program to explore https://www.currentai.org
She made waves by championing smaller language models in a world obsessed with size. Her focus on data quality over quantity is a refreshing counterpoint to the "bigger is better" narrative dominating AI discussions. Pleias is ambitiously building open data assets for training, aiming for a hefty 2 trillion tokens. Their work on a sovereign French model for public services underscores the growing importance of localized AI solutions. Stasenko's cautionary note about the potential disappointment in enterprise AI adoption was particularly striking, suggesting a more measured and realistic view of AI's near-term impact. Her emphasis on "boring but effective" AI solutions might just be the wake-up call the industry needs.
She made waves by championing smaller language models in a world obsessed with size. Her focus on data quality over quantity is a refreshing counterpoint to the "bigger is better" narrative dominating AI discussions. Pleias is ambitiously building open data assets for training, aiming for a hefty 2 trillion tokens. Their work on a sovereign French model for public services underscores the growing importance of localized AI solutions. Stasenko's cautionary note about the potential disappointment in enterprise AI adoption was particularly striking, suggesting a more measured and realistic view of AI's near-term impact. Her emphasis on "boring but effective" AI solutions might just be the wake-up call the industry needs.
