Unpacking the intellectual structure and evolution trend of general-purpose technologies development in innovation studies

General-purpose technologies (GPTs) are crucial for advancing long-term economic growth. Previous research on GPTs has primarily focused on economics. However, in the innovation field, firms face greater challenges in appropriability and value creation due to GPTs' externalities. Research on GPTs in this flexible field may exhibit unique characteristics. Despite growing academic interest, related research remains fragmented, lacking a comprehensive theoretical system. Traditional literature reviews and bibliometric analyses often focus on the most cited articles, leading to citation biases and an emphasis on impact over theme discovery. Combining topic modeling with manual coding allows for the iteration of existing theories and the creation of new theoretical frameworks. Our study analyzed 532 articles on GPTs in the innovation field, identifying 11 topics using the LDA topic model. Through manual coding and the PyLDAvis visualization tool, we identified four research areas: jungle of GPTs, profiting from GPTs innovation, industrial convergence, and economic growth and wage inequality. We examined the evolutionary trajectory, and theoretical architecture of GPTs research, proposing a comprehensive framework. We urge scholars to extend GPTs research from the firm to the ecosystem level, consider the standardization and evolution of next-generation GPTs, and diversify research methods.