Nowadays, artificial intelligence has become an integral part of everyday life, penetrating into various aspects of society and business. The technology is in demand in medicine, finance, education and other fields. However, with this rapid development comes new challenges. Developers and researchers face a number of complex problems that require deep understanding and solutions – we will talk about this with Egger Mielberg, the founder of Arllecta.
Artificial intelligence has become perhaps one of the most discussed topics these days. How would you characterize AI and its key features?
Artificial Intelligence can be categorized as a field of computer science, the model seeks to create intelligent machines capable of performing tasks that normally require human intelligence. Its key features include the ability to learn from data, adapt to new situations, make decisions, recognize patterns and language, and the ability to solve complex problems in real time. These features allow AI to improve performance, optimize resources, and find innovative solutions in
various areas of life and business. However, I should note that not all neural networks that are now positioned as artificial intelligence actually are.
Why? What criteria do they fail to meet?
I emphasize two key features of true artificial intelligence that make it unique. The first is the ability to generate its own new knowledge and adapt to changing conditions in communication with humans. Artificial intelligence must understand and respond to changes in the tone, focus, and subject matter of a conversation. The second characteristic is the property of extracting basic definitions and knowledge from user interactions that can be used in the future. In other words, it can shape its own experience. In my opinion, in most existing models, including GPT chat, these characteristics are still not realized, and this is confirmed by many studies and developers themselves. Neural networks do not have a true understanding of context. They can analyze data, but they cannot truly understand the content or emotional nuance of information, which is an important aspect of human intelligence. That’s why my team and I are actively working on our own tools that could actually create artificial intelligence that can go beyond the current limitations.
What goals do you have for yourself?
I intend to change the paradigm of artificial intelligence and move the technology in the right direction. We aim to show the difference in the point of view and meaning of artificial intelligence by identifying the characteristics that differentiate our solution. Our goal is not just to become industry leaders, but to set new standards and development directions in the field of AI. This is important so that people’s research and efforts are channeled in the right direction rather than drifting towards useless exploration. At the moment we are actively testing our programs and plan to open access to them for users in the near future.
Do you already have tested developments that have shown practical results?
Of course. Let’s start with the fact that back in 2010 I decided to develop a search engine comparable to Google. The model was based on semantic search – the system analyzed the user’s query, identified the main problem and offered relevant solutions. This approach was fundamentally different from that of typical search engines, whose algorithms look for simple matches of words and phrases. I tested the system together with my American colleagues, and the results of these tests confirmed the superiority of my system over Google and Yandex. Some of the algorithms were sold to Yandex, and I continued working in the field of artificial intelligence.
My current projects are related to mathematical models in the field of the Internet of Things and solutions for call centers. In particular, I launched an algorithm to track the activity of workers at large manufacturing plants. Our system detects their state, location and other parameters, and then analyzes and adjusts processes in real time, taking into account their position and speed. We have also developed smart systems for contact centers that process customer requests at the level of meanings, trying to understand their essence and solve issues without transferring them to an operator. Their task is to provide a high level of service without involving human resources.
In addition, I have a project in the field of medicine.
Can you tell us more about it?
We have developed a unique model that allows us to detect relationships that remain invisible when using traditional methods of analysis. I call them “semantic connections” – these are interconnections between objects of different nature. Our system is capable of revealing, say, a connection between an airplane and an art exhibition, which may not be obvious to a researcher. The uniqueness of the approach lies in its ability to reveal hidden connections at the micro level – biochemical and molecular. Let me explain how this works: we can identify the specific source of inflammation in the human body. For example, changes on the skin can be associated with the dysfunction of an organ that is far from the site of inflammation, it is directly caused by the work of a certain part of the brain. Our method allows us to decipher these connections and reveal the hidden causes of diseases.
In what area of medicine do you intend to use this algorithm?
We are working on a diagnostic model that detects oncology at early stages. Unfortunately, existing diagnostic methods show only 60% accuracy in detecting certain cancers. This percentage is incredibly low. In addition, given that the majority of cancers detected are already at stage three and above, this means that the chances of a complete cure are extremely slim. This is why this new approach is so important. Our methodology provides more accurate results in diagnosis and allows us to start treatment as early as possible. And this is a key step in the fight against this terrible disease.
So, we see that AI is gradually taking over more and more areas of our lives. What are the challenges facing researchers and developers in this field?
One of the key challenges is the need to process large amounts of data. As technology advances, data is becoming larger and harder to analyze, so developers have to implement efficient methods to collect, store and analyze this data to extract valuable insights. Cybersecurity and privacy is always a hot topic. As the amount of information used to train algorithms increases, the likelihood of leaks and misuse of this information increases. And of course, ethical issues are on the agenda. The development of AI must take into account the possible negative consequences of its use, such as discrimination, job loss due to automation, and threats to privacy. Researchers should work to create technologies that benefit society while minimizing risks and negative impacts. In short, developers and researchers face many challenges, but their goal remains the same: to create sustainable, safe and ethical artificial intelligence systems that can revolutionize the world, improving our lives and society as a whole.
Featured image credit: Steve Johnson/Unsplash