Unleashing Artificial Intelligence As A General-Purpose Technology

In our era of rapid technological transformation, Artificial Intelligence (AI) stands on the brink of becoming a powerful general-purpose technology (GPT) akin to electricity or the steam engine. These foundational technologies fundamentally reshape industries and redefine society by following an evolutionary trajectory that moves from small improvements to system-level change. History shows us that realizing the full potential of GPTs demands both an understanding of their progressive phases and a forward-thinking mindset, especially to avoid the productivity lags that have plagued previous technological revolutions.

Understanding General-Purpose Technologies and Their Phases

General-purpose technologies share three core features: they are pervasive, continuously improving, and serve as platforms for other inventions and innovations. Technologies like electricity, the steam engine, and the Internet each exhibit these characteristics and have catalyzed waves of invention, innovation and productivity. AI shares these same traits and has the potential to reshape countless sectors, driving advancements across healthcare, logistics, finance, and beyond.

I make a distinction between invention and innovation. General-purpose technologies (GPTs) are significant drivers of invention. GPTs, by nature, create entirely new possibility spaces, which enable the development of breakthroughs that redefine what’s possible across industries and society, establishing entirely new pathways and capabilities that had no prior equivalent. For example, the steam engine led to the invention of mechanized manufacturing and railroads, electricity brought about electric motors and modern communication, and the Internet spurred the creation of digital services, e-commerce, and cloud computing.

While GPTs often lead to innovations — improvements or adaptations that enhance existing processes — they primarily catalyze invention by providing foundational capabilities that make entirely new kinds of solutions possible. As each GPT advances, it continuously expands the horizon for further invention and creativity, building a platform for subsequent innovations that capitalize on these foundational breakthroughs.

As highlighted by Ajay Agrawal and co-authors in Power and Prediction, GPTs typically evolve through four main phases:

1. Current State: Before a GPT is widely adopted, industries operate within existing structures, using established processes that are often manual and lack the efficiency these technologies can provide. In the case of electricity, for example, factories relied on centralized steam engines to power machinery, which required rigid layouts, numerous adjustments, and high maintenance.
2. Point Solutions: The initial use of a GPT tends to focus on isolated tasks or functions that are enhanced for specific gains without fundamentally changing the system. Early adopters of electricity replaced steam engines with electric motors but retained the central drive shaft model, seeing improvements in power reliability and control without structural change.
3. Broader Applications: As the GPT spreads across functions, it starts to create interconnections that improve overall productivity. When electric motors began to power individual machines, factory layouts gained flexibility, improving productivity while still following assembly-line structures.
4. System-Level Change: The final phase sees a fundamental redesign of operations or entire industries. When factories fully embraced electricity, they could eliminate central shafts, decentralize operations, and adopt assembly lines, enabling just-in-time manufacturing and greater scalability. This system-level change redefined manufacturing and set the foundation for modern industry.

AI as a Modern GPT: A Similar Evolutionary Path

AI is poised to follow a similar evolutionary path. Currently, AI often enhances individual tasks like data entry or inventory management, marking the beginning of point solutions. However, as it progresses, we will see AI integrated across functions, enabling organizations to harness predictive analytics, optimize logistics, or coordinate patient care seamlessly. Ultimately, AI’s evolution will culminate in system-level transformations that could redefine industries-from autonomous, real-time supply chains to decentralized healthcare systems.

For example, consider the emergency room of the future. Today, doctors rely on human experience to assess, diagnose, and treat patients. AI is beginning to assist in isolated tasks, like reading X-rays or helping triage patients. As applications expand, AI could predict patient volumes, optimize staff allocation, and link diagnostics with administrative processes. In time, AI might fully transform the ER, shifting urgent care to predictive, at-home solutions, potentially decentralizing emergency services entirely.

Jeremy Rifkin’s concept of a Logistics Internet aligns well with the system-level transformation AI could drive in logistics. In this vision, logistics becomes a fully integrated, digital network where data, AI, and IoT technologies coordinate in real-time to optimize supply chains from production to delivery. Just as AI could shift logistics from linear supply chains to adaptive “supply webs,” Rifkin’s Logistics Internet envisions a dynamic ecosystem that leverages connectivity and automation to create a resilient, decentralized logistics infrastructure. This shift would enable logistics systems to respond to demand instantaneously, minimizing inefficiencies and localizing supply chains.

Learning from the Productivity Paradox

Past GPTs, such as electricity, demonstrate a pattern known as the “productivity paradox,” where it can take decades for productivity gains to materialize. Economist Paul David famously noted that electricity took around 40 years to appear in productivity statistics. Early adopters often used new technology to mimic old processes, limiting its transformative potential until organizations adapted to leverage its unique capabilities. This delay shows that accelerating AI’s productivity potential requires more than task-based improvements.

Accelerating AI’s Progression: A Strategic Approach for Leaders

To ensure AI reaches its transformative potential, leaders should embrace high-level strategies that foster ecosystem-wide collaboration, anticipate systemic change, and cultivate future resilience.

1. Adopt an Ecosystem Mindset: AI’s transformative power lies in ecosystems where diverse stakeholders co-create new forms of value that span organizations and sectors. In these ecosystems, value creation and capture shift from isolated improvements to shared, system-wide advantages. Leaders who embrace an ecosystem mindset can accelerate AI’s progression through its phases by aligning on shared goals, flexible governance, and a collaborative approach to innovation. This ecosystem focus drives AI adoption beyond traditional boundaries, facilitating a path toward system-level change that reshapes entire industries
2. Rehearse the Future: With AI’s path difficult to predict, leaders should conduct structured “future rehearsals,” exploring AI-driven scenarios to understand potential impacts and risks. Future rehearsals provide insights into system-level changes and prepare organizations for adaptability and resilience amid rapid technological shifts.
3. Cultivate a Vision for Systemic Change: Instead of focusing solely on incremental improvements, leaders should develop strategic visions that anticipate the system-level transformations AI can drive. This vision might include redefined business models, new customer experiences, or pioneering markets, ensuring that AI becomes integral to the organization’s core evolution.
4. Foster Cross-Organizational Collaboration: Transformative change requires collaboration across organizations, including stakeholders from various sectors, regulators, and community partners. By creating consortia or alliances, leaders can accelerate AI development in shared areas, establish ethical standards, and build frameworks for interoperability.
5. Define Long-Term Success Metrics: Metrics for AI should reflect broader, long-term goals that go beyond short-term gains. Board leaders should measure adaptability, ecosystem impact, and resilience to ensure AI investments drive meaningful productivity improvements, advancing AI from isolated applications to system-level impact.

Moving Forward: Embracing AI’s Transformative Potential

AI has the potential to revolutionize our economy and society, much like electricity did before it. But realizing AI’s full impact requires a shift in how we approach its integration — focusing on ecosystem collaboration, forward-looking strategy, and readiness for system-level transformation. Leaders who embrace these strategies can foster an environment where AI drives sustainable growth, adaptability, resilience, and the capacity to thrive in a complex, interconnected world. This proactive approach reduces the risk of productivity lag and sets the foundation for enduring value creation across industries, positioning organizations to lead in an AI-driven future.

Originally published at http://frankdiana.net on October 30, 2024.