Bridging the ESG Compliance Gap: How AI-Assisted Reporting Can Scale Proportionate Disclosure for SMEs
University of Nottingham | Nottingham | United Kingdom
Research & Innovation Showcase | 2pm, BST (Find Local Time)
Duration: 30 minutes
Platform: MS Teams
Description: Small and medium-sized enterprises (SMEs) face an increasing burden to produce ESG disclosures under multiple regulatory frameworks (IFRS S1/S2, GRI, VSME). Current guidance assumes large corporate capacity; SMEs lack both the expertise and resources. The project presents research-backed evidence that AI can help democratize ESG compliance without replacing expert judgment.
We showcase three integrated systems built under a British Academy Mid-Career Fellowship:
AI-ESG Platform generates compliant ESG reports across three frameworks from identical company data gathered from diverse sources, addressing the <25% vocabulary overlap challenge: frameworks share such different terminology that simple content recycling is impossible.
Agent-ESG delivers explainable agentic reasoning (planning -> retrieval -> synthesis -> hallucination detection -> adversarial review) to build stakeholder confidence in AI-generated outputs.
AI-Annotation-ESG enables human-in-the-loop evaluation at scale with inter-rater agreement statistics, rigorous quality rubrics, and gold-standard datasets that make AI outputs auditable and defensible to regulators.
Key findings: Validated across ~400 UK SMEs: <3% hallucination rate; mean factual accuracy 87%; 75–90% framework alignment; Cost-effective reports and policy implications for proportionate SME guidance and regulatory clarity.
The project bridges AI capability and policy need, offering regulators and advisory firms evidence-based guidance for responsible AI deployment in compliance-critical workflows.