From NLP to LLMs to agentic AI: The evolution of artificial intelligence
Introduction
Generative artificial intelligence (AI) has permeated global enterprises and societies with a velocity that is without precedent in the history of technology. Yet we must remind ourselves that speed is not a strategy. It is merely a condition. What truly defines this era is the fundamental shift in how these systems have transitioned from calculators of language to actors of intent. We have reached an inflection point with the emergence of agentic AI (AAI) built on fundamental and generative AI building blocks (
Samuel et al. 2024; Acharya et al. 2025; Tripathi et al. 2025).This transition from models that merely respond to those that possess autonomous, tool-integrated architectures demands a rigorous rethinking of our sociotechnical structures. One thing is now certain: governance and account-ability can no longer be treated as administrative afterthoughts or late-stage constraints. They must be designed into the “nervous systems” of future technologies. Where previous efforts settled for the fragmented automation of discrete tasks, agentic systems leverage the triad of autonomy, planning, and memory to command entire processes from end to end. They transform isolated tools into unified agents of performance. However, this has led us to the generative AI paradox: we see widespread adoption, yet measurable enterprise-level impacts remain stubbornly uneven. Although organizations invest and experiment with enthusiasm, transformative returns remain elusive. We have yet to align this new computational capacity with the disciplined practice of management and the rigorous requirements of institutional stability
Read Article
Citation
Samuel, J., Kashyap, R., Misra, V., Tripathi, A., Nguyen, H. D., Daneshmand, M., & Pelaez, A. (2026). From NLP to LLMs to agentic AI: The evolution of artificial intelligence. Journal of Big Data and Artificial Intelligence, 4(1). https://doi.org/10.54116/jbdai.v4i1.82