
Economic classification systems are, by nature, a step behind the frontiers of innovation. The International Standard Industrial Classification (ISIC) system has served the world well—providing a backbone for economic statistics, policy analysis, and international comparisons. Yet, as artificial intelligence (AI), automation, and entirely new business models blur the lines between sectors, the challenge is clear: how can ISIC remain relevant and actionable in an economy that is changing faster than ever before?
A walk through any major city—or a scan of startup incubator portfolios—makes the challenge real. Companies are launching drone-based delivery services, AI-powered legal research, virtual reality fitness studios, and bioengineering firms that didn’t exist even a decade ago. Many such businesses do not fit neatly into existing ISIC categories. They may span multiple codes or defy traditional definitions of manufacturing, services, or distribution altogether. For statisticians, economists, and policymakers who rely on ISIC for clarity, this raises the risk of data blind spots—where new industries are undercounted, misclassified, or invisible in official statistics.
Recognizing this, there is a growing consensus that ISIC must evolve—not just in content, but in the speed and method of its revisions. The traditional approach, based on periodic expert review and international consultation, can lag several years behind market realities. What if, instead, ISIC became a living system—capable of dynamic updates informed by the rapid pace of economic change?
One promising approach is to harness the very technologies that are disrupting industry boundaries in the first place. AI-based text-mining and natural language processing can be deployed to scan company descriptions, patent filings, and business registry data worldwide. By analyzing the language companies use to describe their activities, algorithms can flag clusters of terms or business models that don’t map cleanly to existing ISIC categories. If, for instance, a surge of new registrants describe themselves as “autonomous mobility platforms” or “AI-powered logistics orchestration,” the system could suggest new subcategories or trigger expert review.
Such a feedback loop, blending human judgment with machine learning, would make ISIC more adaptive. Rather than waiting for the next major revision cycle, statisticians could propose interim updates, pilot new codes, or test the relevance of emerging categories in parallel with traditional ones. This would keep the classification system responsive and forward-looking, reducing the lag between economic innovation and statistical visibility.
Another aspect of future-proofing is crosswalks—linkages between ISIC and other evolving classification systems, such as the North American Industry Classification System (NAICS) or bespoke codes used in fintech, biotech, or the digital platform economy. As industries increasingly straddle jurisdictions and regulatory regimes, these crosswalks ensure comparability and support international analysis. When a new sector emerges—say, decentralized finance platforms or AI-based diagnostic tools—statisticians can trace its footprint across multiple systems, flagging discrepancies and harmonizing categories as needed.
A dynamic, tech-enabled ISIC would also support better policy and investment decisions. Governments, for example, could use real-time classification data to spot trends, identify skills gaps, or target incentives for sunrise industries. Investors and researchers could analyze emerging clusters, benchmark innovation, or evaluate sectoral risk with greater confidence. Even regulatory agencies—charged with everything from consumer protection to antitrust enforcement—would benefit from an up-to-date map of economic activity, especially in fast-evolving fields.
However, this vision is not without challenges. AI-based classification must be transparent and accountable—avoiding the introduction of bias or the overfitting of codes to fleeting trends. The human element—expert review, stakeholder consultation, and international agreement—remains essential, especially when proposed categories have policy or regulatory implications. There are also practical questions: how to manage the volume of data, how to decide when a new activity warrants its own code, and how to maintain backward compatibility for longitudinal studies.
Despite these obstacles, the direction is clear. The future of ISIC must be agile, inclusive, and technologically savvy. It should invite input from startups and tech firms, not just legacy industries; it should build partnerships with data scientists and international organizations; it should see classification as an iterative, responsive process, not a fixed periodic event.
Some countries are already piloting these ideas. National statistical agencies are experimenting with AI-based registry scanning and crowd-sourced updates to sector codes. International bodies, recognizing the need for speed, are considering proposals for rolling ISIC amendments—enabling quicker updates between full-scale revisions. The hope is that, over time, the system will become not just a mirror of the past, but a radar for the future.
Adapting ISIC to the realities of AI and emerging industries is less about technology and more about mindset. It’s about building a classification system that is both rigorous and flexible—one that honors its legacy but isn’t afraid to change. If done well, ISIC will remain what it has always been: an indispensable tool for making sense of the economic world, no matter how fast that world evolves.