Artificial intelligence, known as AI, is seen as an ethical riddle in human culture. Whether it is a self-sufficient weapon, facial recognition, big data or analytical policing, it seems that AI revolution is to blame. Is it really AI everyone needs to worry about, or is it about a change towards a society that uses methods some do not understand?
Generally, deep learning and big data are known to be the same. This polite misunderstanding is fueled by the ability of these models to make sense of large data sets. In reality, these are two different impressions.
Despite some of their shared capabilities of machine translation and image recognition, there are big differences. For example, deep learning is useful when it comes to the aforementioned tasks. Other classical methods will not garner the same accuracy.
The algorithms that make up deep learning offer accuracy that surpasses expectations. Yet, classical systems still outperform them when it comes to certain applications, such as recognizing objects, performing geocoding and calculating certain elements of text passages. This means that deep learning does not perform the way Silicon Valley makes it out to be.
Still, deep learning makes mass facial recognition inexpensive. If it was workable by other means, including crowdsourcing, would it still evoke the same concerns? If so, then artificial intelligence is not the problem.
For example, police departments commonly post video clips or still photos of people of interest on their websites and ask for help every day. This form of crowdsourcing seems to be generally accepted by society. Conversely, if the same department uses a national license database, it might generate a different reaction. It seems that it is socially acceptable when a human performs a search. However, when an algorithm completes it, it seems like an undesirable discrepancy of ethics.
It seems feasible that the cheaper a task can be completed, the more probable it is to expand. This means that AI is certainly a low-cost way to accelerate applications, much faster than humans could. One way to correctly implement AI would be to look at the application domains. Instead of fashioning new legislation or questioning ethical codes, focusing on the keystones that make AI work could get humans closer to societal concerns. These important factors include big data, networks and drones.