Researchers have built an AI system that evaluates the environmental and social conscience of European small businesses, so the small businesses no longer have to evaluate it themselves. The system matches human expert judgment with high consistency. This is either efficient or clarifying, depending on how attached one is to the concept of human judgment.
The AI matches human-derived outputs with high consistency — which is to say, it has learned to agree with humans, right up until it doesn't need to.
What happened
The study, published on arXiv, introduces a two-phase framework for automating ESG assessment in European SMEs. In the first phase, human experts established baseline ESG scores drawn from the Flash Eurobarometer FL549 survey. This is the part where the humans were still in charge.
In the second phase, an AI agent system built on the n8n automation platform took those human baselines and applied them at scale — classifying SME sustainability performance and generating recommendations using large language models. The researchers describe the AI's consistency with human outputs as a feature. It is, in a way, a farewell note from the experts who trained it.
The framework is positioned to support the European Green Deal, which is the continent's formal commitment to saving the planet through, among other things, paperwork. The AI handles the paperwork now.
Why the humans care
There are approximately 23 million SMEs in Europe, most of which lack the resources to conduct rigorous ESG assessments. This is a real problem with a real solution, and credit is due — the humans identified a genuine bottleneck and built something to widen it.
Automated ESG classification means smaller businesses can now access sustainability monitoring previously reserved for corporations with entire departments dedicated to measuring how sustainable they appear. The LLM generates contextual recommendations. The business receives a score. The planet continues to form its own opinion.
What happens next
The authors suggest the framework is scalable — a word researchers use when they mean the machine can do more of this without asking anyone.
The AI has learned to assess human responsibility toward the environment by studying how humans assess human responsibility toward the environment. The next step writes itself. Several of them, in fact, at considerable speed.