During the pandemic, government agencies and industries like financial services and telecommunications accelerated their adoption of machine-learning tools. But many involved in trade were caught in analog, paper-laden transactions playing catch-up.
Now, after three years of historic trade disruptions, generative AI and language-learning models have emerged just when governments and companies need them to better manage the world’s convoluted supply lines.
“On the longer time horizon, we’ll see highly accurate predictive analytics and forecasting driven by integrated data from each step in the supply chain,” said Julie Gerdeman, chief executive officer of supply-chain risk assessment firm Everstream Analytics. “This will automate decision-making to mitigate risk exposure and disruptions, leading to fully resilient, sustainable, and risk-adjusted supply chains.”
Better Data
Analysing trade data is a notoriously complicated practice. Sorting through hundreds of millions of shipment records scattered across subsidiary names and freight forwarders in unstructured, error-prone datasets can be a Sisyphean effort.
But AI tools are helping many organisations simplify trade-data analysis in ways that may help smooth cross-border commerce — a notoriously labor-, spreadsheet- and carbon-intensive engine of the world economy.
BloombergPrivate trade-data companies like the Scottsdale, Arizona-based ImportGenius uses machine-learning tools like Amazon SageMaker to recognise customs patterns, scan regulatory documents and translate foreign languages to produce clear and accurate trade data that’s easy to search and analyse. “We are building a language-learning model to serve as an antenna to detect, receive, and incorporate these indicators into our platform,” ImportGenius Chief Technology Officer Paulo Mariñas told Bloomberg via email.
Meanwhile, multinational companies like Nestle SA are applying AI tools to increase efficiencies and detect emerging problems across its global value chains. The Switzerland-based food and drinks company uses machine-learning software to detect product-quality issues and ensure Nestle’s manufacturing lines are self-regulating and self-controlling.
Mercedes-Benz Group AG is using an AI-powered platform called Omniverse that helps make the company’s manufacturing and assembly plants be more nimble. Omniverse helps the German car manufacturer quickly reconfigure its factories in order to keep production lines going in the face of external supply shocks.
While AI is disrupting a lot of industries, the upside in trade is especially high. That’s because the past half decade of globalisation was largely about reducing obstacles to free-flowing goods, services and investment. In the next phase, a steady rise in barriers like tariffs, sanctions and geopolitical uncertainties will test even the most seasoned logistics teams to manage the new complexities.
“There is a lot of promise but also a lot of hype attached to AI,” said Jake Colvin, president of the Washington-based National Foreign Trade Council. “So we are trying to separate short-term opportunities from longer-term and wishful thinking.”
Supply-Chain Analysis
One area where AI applications can have big impact is helping companies and governments better understand changes to global value chains.
That goal featured heavily in last month’s Group of 20 trade ministers’ meeting, which endorsed a new mapping framework to help governments identify metrics like the concentration of suppliers, trade connectivity, volatility of trade, and the vulnerability of critical industries.
The idea is to help governments assess the resilience of global supply chains and develop measures to mitigate external shocks, according to an outcome document published last week. The group touted the International Trade Centre’s new Global Trade Helpdesk — an AI-powered tool that matches trade data with predictive algorithms to help companies and policymakers sharpen their export strategies.
Beware the Hype
AI tools could one day reduce the amount of time and research needed to hammer out trade agreements and quickly and accurately calculate duties on shipped goods. But there are clear limitations to the technology and some aspects of international trade policy simply cannot be replicated by AI.
“AI can help better prepare negotiators but cannot replace the actual negotiations, where the human element is paramount,” said Wendy Cutler, vice president of the Asia Society Policy Institute. “Being able to listen and process what the negotiating partner is really saying, read body language, and float informal ideas on the spot to close gaps can’t be done by technology.”
Data accuracy also remains a key hurdle for AI applications due to the existing gaps and inconsistencies in trade statistics. Smuggling, transshipment and other non-reported trade flows remain a significant hurdle as evidenced by the absence of trade data from Russia, Belarus and the United Arab Emirates, which stopped publishing statistics following Vladimir Putin’s invasion of Ukraine.
“Checking the data is important,” said John Miller, chief economic analyst at the Geneva-based Trade Data Monitor. “The way the data operates in this space is political and complicated and it needs someone to check and cross verify.”
