How will artificial intelligence affect our daily lives in the future? Where is the development of artificial intelligence within the Faculty of Engineering? We talked to István Erdei Timotei, one of the participants of the NeurotechEU project, a teaching assistant at the Department of Air- and Road Vehicles of the University of Debrecen, Faculty of Engineering.
How does artificial intelligence appear in training?
In the NeurotechEU project, we are presented with industrial robotics. Artificial intelligence is very difficult to define. ChatGPT, for example, focuses on communication: it translates the text or gathers information about the question. There are also many other AI systems, each with different advantages and disadvantages.
For example, ChatGPT AI, which is best known in our country, can translate beautifully into Hungarian, which was one of the reasons why it is so widely used in Hungary. Its integration into education, on the other hand, is a more serious issue, because even routine engineering tasks can be automated with it, but for more complex designs, or when something related to interaction is necessary – for example, something needs to be controlled in space – it is still in its infancy.
We started to develop one of our labs, where we also use so-called additive manufacturing technology and robotic units for the vehicle industry, around 2012 with Dean Géza Husi in accordance with industry trends. If we were to ask artificial intelligence about exactly what developments it would recommend as the next step in relation to such a lab, it is not certain that it would be able to fully define it.
Artificial intelligence is not infallible, many things affect its operation. They can give ideas, but they cannot bridge, for example, compatibility problems between machines, at least for the time being, they still have to be solved by the engineers.
It should be seen that AI applied in many other fields is also under development. There are AI systems whose main task is to generate images. There, a lot depends on how we formulate the individual words that define the image we want to generate. If the wording is imprecise, there will be a discrepancy, and it is not certain that we will get the expected result. This area is also important to us because we use image analysis techniques during design and production. And the data set to be used for teaching AI can be produced in several ways. The necessary data set can even be prepared from moving image material so that individual frames can be converted to JPG-PNG image format. However, those who are experienced in making video materials or photographs know that a lot depends on the preparations (lighting, focus, etc.) and that we do not always get consistent quality.
The possible way is to create the data set virtually from the given 3D CAD model and use CGI rendering to produce the images for image analysis. In this case, we set the resolution, texture, lighting, etc. (Render means an image generated by the computer, based on some algorithm or model, with the help of it.)
However, we also use artificial intelligence in other cases. We work with industrial robot units in the NeurotechEU topic. We must use the available resources optimally, and artificial intelligence provides effective help for this by shortening the processing of our data to be evaluated.
In education, we draw students’ attention to the fact that artificial intelligence is now available to anyone, but engineering thinking is still very important and a creative approach to the problem to be solved must be taken. And don’t rely so much on the comfort provided by technology, as this can reduce their independence over time.
Like using the phone…
Exactly, and in this regard, it can be observed that things have become very impulse-based. It is much more difficult to capture and maintain attention in education, and digital content consumption habits have also contributed to this in recent years. Furthermore, entertainment electronics also play a significant role in this, so it is a relatively big challenge for us, educators, to organize teams that can then actively participate in research in the longer term. On the part of the students, there is a strong curiosity about artificial intelligence, but many of them are distrustful or afraid of it because it brings change. Among other things, it will have a big impact on the labor market.
Can the number of virtual classrooms increase in the future?
Yes, more virtual classrooms may open in the future. We have prepared a program in this regard, and from here it is up to us how much we actually want to include this in education. During Covid-19, we also used it during events and it proved to be functional. Now the question is how much we want to maintain this state. I think some parts of it will remain online, but it will also have physical elements. Robot cells and machine handling, for example, cannot be completely transferred to virtuality, because physical control is needed so that the student can feel and see what it means to operate machines and what it entails. For example, a specific production line can be modeled in great detail virtually, but when it comes to going out to program or maintain the site, we feel that more is needed to transfer this kind of knowledge.
Where is the development of artificial intelligence within the Faculty of Economics?
In-house, we actively use neural networks to teach the systems. The development of material handling and sorting robots is also important. In the case of the latter, it is important to be able to detect the object in addition to selection based on color and shape. For example, be able to determine whether the object being examined is damaged. But we also have a system – we call it Emma – that can talk to us. For example, it can warn you if someone wants to enter the robot’s workspace.
It should also be taken into account that humanoid-type robots are more difficult to accept, even by the younger age group. This will certainly change over time, but we also paid attention to this in connection with our individual developments. Faceless systems are more easily accepted.
What is the reason why robots with faces have become popular in Japan, South Korea, and other Asian countries?
The environment is different. If we go a little deeper into the subject, we can see that the android-type robot has been a well-known concept there for a very long time, and even as a child, meeting this type of machine is no longer considered a curiosity or a defining part of the culture. For this reason, the social acceptance of this type of robot system is greater. Also, the aging society has a fundamental interest in the use of such machines so that they can help them in the future. Sooner or later, this process will also take place in other countries, so the acceptance of such machines will improve in our country as well.
When will it be possible to get an exoskeleton, i.e. an artificial skeleton that helps people move, in stores?
For the time being, the aging of society in our country is not as dramatic as in some East Asian countries, so the wider distribution of exoskeletons in our country is still the music of the future. By the way, they can also be used very well in the military industry, which is known to drag other industries along. But, it can be a great help in production, as it can speed up the execution of storage tasks.
In many cases, as we can see, technology is already a given, the only question is when and how we want to use it.
– Sándor N. Nagy –