A General Purpose Technology (GPT) is an innovation that can be applied across multiple sectors of an economy and generates sustained productivity improvements over extended periods. Unlike sector-specific technologies that enhance efficiency within particular industries, GPTs fundamentally restructure economic organization and create cascading waves of innovation across diverse applications. Historical examples include the steam engine, electricity, and the internal combustion engine—each triggering widespread economic reorganization and decades of technological adaptation.

The concept of General Purpose Technology emerged from economic research examining long-term productivity patterns and technological diffusion. David (1990) established the foundational framework, identifying key characteristics that distinguish GPTs from ordinary technological improvements ¹⁾.

GPTs demonstrate several defining attributes: they exhibit high initial productivity relative to predecessor technologies, possess substantial room for improvement and elaboration, generate widespread spillover effects across industries, and require complementary innovations to realize their full economic potential. The steam engine, for instance, required development of new materials, safety mechanisms, training methodologies, and industrial organizational structures before delivering maximum economic benefit.

The adoption curve for GPTs typically spans decades, with initial phases involving technological consolidation, infrastructure development, and organizational learning. This extended timeline reflects the complex interdependencies between technological capability and institutional adaptation required for economy-wide restructuring.

Economist Carlota Perez developed an influential model describing how General Purpose Technologies unfold through distinct phases, each characterized by specific economic dynamics ²⁾. Her framework identifies two primary phases following a GPT's emergence:

The Installation Phase represents the period of turbulent transformation where new technological possibilities clash with existing institutional structures. This phase features rapid innovation, financial speculation, infrastructure investment, and significant displacement of established practices. Incumbent industries resist change, creating conflict between old and new economic paradigms. The installation phase typically generates asset bubbles as investors extrapolate technological potential beyond immediate economic reality.

The Deployment Phase begins following financial correction, when complementary technologies, institutional frameworks, and organizational practices mature sufficiently to leverage the GPT's full productive capacity. This phase demonstrates sustained productivity gains, expanded employment, and broad-based economic growth. The transition between phases represents a critical juncture where speculative excess gives way to pragmatic implementation.

Contemporary economic analysis increasingly treats artificial intelligence as a General Purpose Technology comparable to electricity or computing infrastructure ³⁾. AI demonstrates the fundamental GPT characteristics: broad applicability across sectors (healthcare, manufacturing, finance, education, creative industries), substantial potential for productivity enhancement, requirements for significant complementary innovations (data infrastructure, workforce retraining, new business models), and multi-decade implementation timelines.

The current period appears to correspond to Perez's installation phase. Large language models and other AI systems have generated extraordinary investor enthusiasm, substantial capital allocation, and ambitious claims regarding economic transformation. Simultaneously, significant institutional challenges persist: regulatory uncertainty, workforce displacement concerns, questions regarding complementary infrastructure requirements (computational capacity, data governance), and the need for novel organizational practices to integrate AI effectively.

This installation phase encompasses both remarkable technical progress and genuine challenges. The tension between transformative potential and implementation barriers characterizes this period of turbulent innovation. Financial markets have reflected both genuine technological advancement and speculative excess, as is typical during GPT installation phases.

The productivity implications of treating AI as a GPT are substantial. Historical GPT cycles have generated 2-3% additional annual productivity growth during deployment phases, with compounding effects extending across decades ⁴⁾. However, realizing these gains requires successful navigation of the installation phase, during which productivity may actually decline in sectors undergoing transformation due to transition costs and implementation inefficiencies.

Complementary investments prove essential for GPT value realization. During electricity's deployment phase, factories required complete architectural redesigns to distribute power effectively—the technology alone provided minimal benefit without organizational adaptation. Similarly, AI deployment requires investments in worker retraining, process redesign, data infrastructure, and new business model development. Economies or firms that successfully coordinate these complementary investments achieve disproportionate gains.

The distributional effects of GPT transitions deserve consideration. While aggregate productivity may improve substantially during deployment phases, the installation phase frequently creates concentrated benefits and widely dispersed costs. Workers in declining sectors experience displacement before new opportunities emerge elsewhere. Geographic regions dependent on disrupted industries face extended adjustment periods. The political economy of managing this transition period significantly influences both the speed of deployment and the sustainability of the benefits.