.png)
Loading...
16 March 2026 / Essay
Email
Linkedin
Facebook
X
Copy LinkAI has moved from a technical marvel to a board-level priority, raising questions about productivity, labor displacement, and long-term growth. As organizations race to deploy generative AI, leaders face familiar questions: Where will it create durable value? What will limit scale? And how do we avoid overinvesting in capabilities that fail to translate into performance?
This essay revisits the expert systems boom—an earlier wave of applied AI—of the 1980s to draw lessons for today’s business leaders. It examines what drove adoption, what limited scale, and why enthusiasm faded. It then compares those constraints with today’s generative AI systems, which shift the bottleneck from explicit rule encoding toward probabilistic modeling—bringing new opportunities but also new operational, governance, and workforce challenges.
A previous wave of applied AI, expert systems, delivered real productivity gains across multiple industries but their impact proved bounded rather than transformative.
Adoption (or lack thereof) was organizational as much as technical: At the firm level, user acceptance, governance, maintenance burden, and strategic alignment ultimately determined whether expert systems delivered value, often more than the raw technical capabilities of the systems themselves.
An epistemological bottleneck: AI can outperform individuals in structured tasks but fully replacing labor requires sensitivity to tacit knowledge embedded in habits, context, and judgment, which remains difficult to model.
Three years ago, breakthroughs in data-driven AI systems pushed AI from research labs to the C-Suite. Working models demonstrated
myTCB® Members get exclusive access to webcasts, publications, data and analysis, plus discounts to events.
