AI in International Trade: Successes & Challenges
summary
Global trade is framed as 80–90% classification problem where AI automates routine tariff, export-control and sanctions decisions but struggles with the ~20% ambiguous cases; a concrete legal conflict (German Customs 2022 vs EU 2023 on hair-and-body-wash) exemplifies why language models must be combined with deterministic rules. Implementation uses a hybrid model with rules engine and constraint solver, smaller focused LLMs, many more short prompts (compute up by ~10–100×), QA and plausibility checks, and an argumentative AI approach that flags low-confidence cases for human review.
AI-written summary · about
chapters (9)
- 0:00AI in global trade
- 1:14Human intuition challenges
- 3:48Product classification experiment
- 7:52Customs classification subjectivity
- 10:22Resolving legal conflicts with AI
- 12:53Trust in AI classification
- 15:05Hybrid AI for classification
- 18:43Automated classification benefits
- 21:22Q&A
related talks
44mAI, Corporate Responsibility & Democratic Legitimacy
46mAI for Legal : The story of hype, adoption and what comes next
32mAI in digital accessibility: challenges, benefits and risks
16mNavigating the AI Jungle
01h 01mPanel discussion: Practical and responsible applications of AI in businesses
33mAI is coming for your job
22mAI and Blockchain Shaping The Future
46mAI vs Humans Rewriting the Rules of Success