At first, we were able to mimic simple language learning. Of course, it was only by text, and eventually they just got stuck on a loop. Pretty easy to figure out if you were talking to an AI or not.
Now, we've got physical learning. We load on servo-connections and begin with a basic calibration of the synchros. Within a year, AI-noids are capable of learning their physical limits. Then we give them basic tasks. After about... Gosh, 25 attempts, we realized we needed to code in visual cues for how a task has failed. But since then it's been great!
Advanced robotics over the last 250 cycles has lead to great strides in machine learning and automation. In the last 35 cycles however, Dr. Imahara has helped bring Artificial Intelligence to a symbolence of life.
AI-noids, sometimes referred to as noids, are robots that have undergone rigorous artificial and machine learning, allowing any model to learn and perform tasks within the confines of it's physical design and limitations. The most common types are wheeled or quadcopter designs for survey, reconnaissance, or cleaning. More advanced designs mimic their creators physiology, but may only be capable of five or six individual task types. Amature robotocists take pride in designing AI-noids capable of assisting them or their families in mundane tasks.
'While getting a scalp massage' - love it!