Green Generative AI – 2nd Green AI webinar of ENFIELD

Date: February 4^th, 2026, 14:00-15:00 CET
Location: Online (Teams)

The Green AI pillar of the ENFIELD project invites you to a webinar where we dive into state-of-the-art research on generative AI from a sustainability perspective.

While chatbot conversations dominate the public’s use of generative AI, there are several other forms of this technology that create challenges for environmental sustainability. In this webinar we explore two particular aspects: Image generation and LLMs for software engineering.

Using AI to produce images is known to be particularly computationally expensive, but it remains less studied than text generation. Considering the plethora of model architectures and hardware platforms used for deployment, how does the energy usage behave, and is it affected by particular factors? Is it possible to predict the energy consumption to improve resource planning and usage?

Meanwhile, LLMs have seen large efficiency gains over the last few years, leading to a low energy-per-token cost, but the use of AI agents – particularly in the realm of software engineering – has exploded, significantly increasing the number of tokens being generated. AI coding assistants are used not only through simple prompts and responses, but they use iterative agentic loops to complete their tasks. Coupled with increasingly large context sizes, the token usage is driven to new heights. At the same time we are lacking robust tools for assessing their performance. How do we tackle these challenges while still benefitting from the new technology?

Three experienced research scientists have been invited to shed lights on these matters with the following talks:

Energy Scaling Laws for Diffusion Models: Quantifying Compute in Image Generation, Boris Ruf (AXA)

Rigour, Reproducibility, and Responsibility: Rethinking How We Study LLMs in Software Engineering, Max Hort (Simula)

Agentic LLMs for Software Engineering Tasks: Do They Help or Hurt?, Merve Astekin (SINTEF)

Join us to learn how European scientists are spear-heading the research on sustainable and trustworthy AI.

Participation is free.

The webinar is organized in the framework of the Horizon Project ENFIELD - European Lighthouse to Manifest Trustworthy and Green AI – and it will take place on the Teams platform. A Teams-link will be provided to registered participants on the day before the event.

Secure your spot here (Registration Closed).

Registration Closed

Target audience

The target audience includes young researchers, AI developers & data scientists, AI professionals, members of Digital Innovation Hubs, participants from related initiatives and projects, and anyone interested in AI Agents and their interaction with Humans.

Speakers

Boris Ruf, AI Research Scientist Lead, AXA, France.
Max Hort, Research Fellow at Simula Research Laboratory, Norway.
Merve Astekin, Research Scientist, SINTEF Digital, Norway.

Program

This webinar features a series of talks focused on the development and challenges of artificial intelligence (AI), with a particular emphasis on trustworthy and sustainable AI approaches, agent interactions, and multi-agent systems. Below is the detailed schedule of presentations and discussions with renowned international experts in the field:

Date 4 February 2026, 14:00-15:00 CET	Time	Presentation title
	14:00 – 14:20	*Energy Scaling Laws for Diffusion Models: Quantifying Compute in Image Generation* Boris Ruf, AI Research Scientist Lead, AXA, France.
	14:20 – 14:40	Rigour, Reproducibility, and Responsibility: Rethinking How We Study LLMs in Software Engineering Max Hort, Research Fellow at Simula Research Laboratory, Norway.
	14:40 – 15:00	Agentic LLMs for Software Engineering Tasks: Do They Help or Hurt? Merve Astekin, Research Scientist, SINTEF Digital, Norway.