Artificial intelligence: personal experience, current state and future trends

– Can you please define what is AI in your opinion?

There are a large number of definitions for AI. For me personally, AI is a collection of mathematical algorithms, methodologies and models in combination with programming and technical means developed and constructed by people for execution of specific tasks. First of all, these are the tasks that demand mental human capabilities, such as learning, generalization of received knowledge, recognition of patterns, planning, decision making, communication.

– What is distinction between artificial and natural intelligence?

AI is based on algorithms and software solutions, whereas natural intelligence is based on cognitive processes. There is a little to no doubt left that AI has already surpassed natural (human) intelligence based on such criteria as processing speed, versatility and solution quality. However, I personally believe if we would consider humour and higher-order thinking, human intelligence is still far ahead of AI.

– How AI algorithms have affected your professional area? How significant is this effect and how does it manifest?

Together with my colleagues we have been developing machine vision systems for more than 25 years. The first neural networks in form of Perceptron algorithms have been programmed to recognise vehicle number plates at the end of last century. To note, these algorithms have successfully been a foundation for our commercial software product for more than a decade. For addressing tasks of visual quality control of manufactured products in 2000s we have successfully been implementing so-called classical methods and algorithms for image analysis and shape recognition.

The emergence of convolutional neural networks and deep learning technologies have created new opportunities both for us and our clients. Many of tasks that previously were not addressable now can be resolved quickly and efficiently. These are the tasks for quality control and tracking of raw materials and manufactured items, control of technological operations and condition of industrial equipment, diagnostic of violations of occupational health and safety rules and regulations by personnel. All of these tasks are addressed in the complex industrial environments.

The latest neural networks technologies also have propelled previously developed computer vision systems by bringing them to a new quality level. Despite this, however, classical deterministic algorithms for image analysis still remain relevant. For instance, tasks of data aggregation for distributed multi-camera computer vision systems are still addressed with deterministic algorithms.

– Do you believe that clients are ready for the implementation of solutions based on AI and why? What products based on AI in your opinion are more in demand?

There are a lot of medium and large industries that are inclined to implement or already implementing solutions based on AI. Currently there is an increasing amount of information that indicates advantages of such systems, with AI-embedded solutions becoming more accessible. Systems that are most in demand in the industry are quality control solutions, planning and manufacturing management platforms, and video analytics. There is large interest for medical diagnostic solutions, voiced assistance, forecasting systems, autonomous transportation means and robotised solutions created with implementation of AI.

– What technologies based on artificial intelligence during last several years have impressed you the most? What are these innovations based on?

From the most notable – it is development in the area of image processing and classification. Thanks to the use of deep neural networks, we can recognise objects from the photos with an impressive accuracy. This has impacted the medicine, unmanned means of transportation, analytics for manufacturing and some other areas the most.

Also, it should be noted that there is astonishing development taking place in the field of voice recognition. Today we have virtual assistants that are not only capable to understand and analyse our speech but also provide responses and perform tasks. This substantially simplifies some of the everyday activities and has already impacted such areas as, for instance, client support and data management services.

– Can you please make a prediction for trends in the area of AI for the nearest future?

Probably, we would see the following trends:

• Development and wide-scale implementation of autonomous and robotised systems.
• Large breakthrough, even a revolution, in the area of automated machine learning.
• Sharp increase in variety of successfully solved tasks in the area of situational video analytics.
• Continued active development in the areas of biometrics and biometric data processing.
• Widescale implementation in the medicine, finance and transportation.
• Creation of No-Code and Low-Code tools that enable a wider audience to create, train and utilise hardware-software systems with embedded AI elements.

– Can you please compare the state of development and implementationof AI in your industry and other areas both domestically and worldwide? In your opinion, what region or what country is a leader in development and implementation of AI?

When talking about fundamental AI research, then unfortunately based on the number of publications, the contribution of our domestic researchers is not in a leading position. At the same time, however, if we would consider practical application of AI, then I would say that our country takes one of the leading positions. I believe, this is especially the case for the industrial video analytics, biometric systems, technologies for voice analysis and automated translation.

However, both domestically and worldwide the state of AI development and implementation across various areas of application is not evenly distributed. The utilisation of AI is actively growing and not all possibilities have been researched and studied yet.

At the same time, however, the anticipated opportunities and impact of AI are huge. According to António Guterres, Secretary-General of the United Nations, based on several expert assessments AI can contribute between 10 to 15 trillion of USD by 2030 to the world economy.

As a result, we see large technological companies in the US, China and other countries invest large amounts into the research and development of AI. Unfortunately, domestically we only see a fraction of such investments being made.

Published in October-November 2023 issue of "Safety Systems" journal