Will Artificial Intelligence some day become a problem to humanity
after learning human behaviors and characteristics?
It can be answered in both ways, I think.
Yes, they may become a problem.
With the increasing integration of loads of apps and smart devices in our life, almost everything defining an individual human being is digitalised. For instance, our fingerprints, voice, facial image etc. Apart from these data, we use those apps and devices to track our health (heart rate, calorie intake etc), to plan our schedules, and most importantly to communicate. If some sort of AI engine is integrated into a chat application, for instance, it can learn our typing patterns, conversation style, and hundreds of other unforeseen parameters. Imagine what can be learned about a person if such AI is coded inside every device and every app in your day-to-day use.
We use smart devices and apps to harness their functionalities and features which ease our way of life, and we give them, unintentionally, our identity and sometimes, even our personality. For them, these are the parameters that can be input to some machine-learning algorithm and predict what we will do the next day, or what will happen to us the next day.
This sounds like a major problem, especially when these technologies are indispensable.
No, they may not become a problem.
Humans are really complex creatures and possess the most advanced intelligence technology called the brain. I think the brain can be thought of as a technology in this context. There are several tissues inside the brain that can learn to do certain things themselves. It was proven in a research that some tissues have the capability to perform the functions of other tissues using neuro-rewiring techniques. Imagine that the same tissue that has helped you see until now can be made to help you hear instead. Now imagine mimicking such a technology.
While it is not impossible, for an AI to achieve the brilliance of a human brain is a topic of ongoing research. To train a machine for the purpose, we would have to feed it with gazillion behaviors and characteristics, which it may not be able to handle! After learning some behavior and characteristics, the AI would be said to be smart, but it would still predate us.
So, they may not become a problem at all because of our brain.
The algorithms (like recommender systems) used by Amazon and Facebook influence us or even manipulate us. But, that manipulation is either very obvious (like viewing promoted products) or is in company's best interests (like viewing a certain news piece). It may be even possible that several external parameters are used by these systems to improve your experience. For instance, Google ads show us what we were looking for on an online store when we visit any random website. In most cases, what you see is a result of what you were looking for before. If any attempt to influence does happen, we may learn to avoid it through careful observation or even experience.