Sajitha Sana P

https://doi.org/10.64823/ijter.2621026

Artificial general intelligence (AGI) describes a form of artificial intelligence capable of handling diverse intellectual tasks with human-like adaptability and proficiency. Unlike narrow AI, which is built for specific functions, AGI stands out due to its ability to exercise sound judgment, adapt across domains, and manage a wide variety of challenges. As AI becomes more embedded in everyday life, achieving AGI represents a pivotal advancement in technology. Yet, significant theoretical and technical obstacles remain. This paper reviews existing literature to outline the primary hurdles in AGI research, including limited common sense reasoning, struggles with rare or unexpected scenarios, issues with generating inaccurate information, constrained memory, and weaknesses in reasoning, planning, and lifelong learning. The analysis suggests that progress may come from hybrid models, embodied cognitive approaches, and improved reasoning mechanisms. Improving generalization, computational scalability, robustness, explainability, verification, and safety will also be essential for developing reliable and adaptable AGI systems. Overall, the study presents a detailed examination of current technical barriers and helps shape future research directions in AGI development.