BEYOND GENERATIVE AI: DEVELOPING AUTONOMOUS INTELLIGENT SYSTEMS WITH REASONING, SELF-ASSESSMENT, AND ADAPTIVE DECISION-MAKING

1Hafsa Hussain; Irfan Hanif; Noor Fatima

Authors

1Hafsa Hussain
Irfan Hanif
Noor Fatima

Abstract

The development of artificial intelligence has increased interest in creating autonomous and intelligent systems beyond generative AI. The effects of reasoning capability, self-assessment, and adaptive decision-making on the development of autonomous intelligent systems were studied. This study used a quantitative, cross-sectional survey design. A structured questionnaire using a five-point Likert scale was administered to 220 respondents, including AI researchers, software engineers, data scientists, university faculty members, postgraduate students, and IT professionals. The data were analyzed using SPSS Version 29, including descriptive statistics, Cronbach's alpha, Pearson's correlation, and multiple regression. The reliability analysis showed excellent internal consistency, with an overall Cronbach's alpha value of 0.895. The correlation analysis revealed a high positive correlation between reasoning ability and autonomous intelligent systems (r = 0.79), self-assessment and autonomous intelligent systems (r = 0.75), and adaptive decision-making and autonomous intelligent systems (r = 0.82, p < 0.01). The model explained 74.7% of the variance in autonomous intelligent systems with an R² value of 0.747. Adaptive decision-making emerged as the strongest predictor (β = 0.396, p < 0.001), followed by reasoning capability (β = 0.321, p < 0.001) and self-assessment (β = 0.279, p < 0.001). The results suggest that improving reasoning, reflective evaluation, and adaptive decision-making could be associated with greater autonomy, reliability, and effectiveness of AI systems in the future. The study provides practical recommendations for researchers, developers, and policymakers to support the development of reliable, human-centered AIS.

Keywords : Adaptive decision-making, Autonomous intelligent systems, Generative artificial intelligence, Reasoning capability, Self-assessment, Trustworthy artificial intelligence.