Autism is a neurodevelopmental illness marked by restricted and repetitive behavioral patterns, poor verbal and nonverbal communication, and impaired social interaction. There is not a particular medical test to identify this disorder’s presence and behavior. For appropriate care of ASD and early intervention, an early diagnosis is essential. A person’s conduct and developmental stage can be used to diagnose. Machine learning and deep learning techniques are utilized to predict autism early on, reducing the time required for this process. This study aims to present a thorough review of earlier studies on autism diagnosis, emphasizing the use of deep learning and machine learning and the comparison of their respective results. According to the instruments employed in categorization, this study found that the diagnosis can be divided into categories based on the following: (brain pictures including MRI and EEG, face images comprising facial characteristics and facial expression, eye tracking, and other ways).