Page 121 - Azerbaijan State University of Economics

THE JOURNAL OF ECONOMIC SCIENCES: THEORY AND PRACTICE, V.77, # 1, 2020, pp. 113-132

It is certainly worth noting and keeping in mind that the AI is not even advanced
enough to be able to learn independently but rather is given the command to act. AI
is not capable of transmitting emotions, and it scares humanity significantly. It is
unable to perform functions and tasks that include any emotion (Martinez and
Fernandez-Rodriguez, 2015). Lack of emotion can even be destructive and lead to a
damaging situation. Also, it is typical that the AI employed can take risks and is
unable to recognize the difference between positives and negatives. They also lack
design ability because they lack consciousness and creativity. Although in some
cases an AI solution may be creative, it is minimal, as creativity is less predictable
and expectable and cannot be achieved by using and studying previous data. In the
absence of past data, the AI is unable to get started or make a good decision because
it is not trained in advance. AI is also incapable of transferring knowledge and, in
novel cases, is unable to apply and use logic and ingenuity (Turan et al., 2018). AI
systems can be disrupted even by a situation they have never encountered before.
Because not all AI systems understand causation, they are not always able to
connect events either. Sometimes it is necessary to re-teach how to do a task because
a variable has changed. At this point, the AI may lose its existing expertise that was
associated with the original task (Bergstein, 2020).

Singh (2017) estimates that one of the biggest problems is that the concept of AI is
often confused with simulating a person with software that is equivalent to
intelligence. The AI application is specific and does not necessarily correspond to
the concept of simulation. Years ago, with the advent of 'big data', people were sadly
unaware of the concept of real space. The effect of this extends to AI. The problem
is that most of the data is unlabeled, so it does not mean quality, undistorted data.
Unlabeled data is ubiquitous and will continue to be a major challenge. There are not
so many scientific researchers, professionals, data scientists who can keep up with
the spread of AI. In conclusion, there are no previous research results that could be
effectively used for improvements. The models so far have only limited capabilities,
so the capabilities of the appropriate AI implementations are not fully developed. It
is still an untapped area because it requires a lot of additional research work. There
is a broad context in the area of AI, and to be successful, all components and
elements must be well defined and specified. This is still incomplete today.

Economic effects of AI in Hungary
Automation contributes to GDP growth in Hungary, as it increases productivity
levels and also addresses labor shortages in certain areas (PWC, 2019). While GDP
in Hungary declined after 1989 due to the transition, there was an increase between
1995 and 2009 as a result of improved capital investment and productivity. After the

121