For good or evil purposes?

Neural networks have already entangled our privacy. What will happen next?

Global technological innovations inevitably bring us closer to the moment when artificial intelligence (AI) will become part of everyday life for everyone without exception. According to the prediction of the famous British physicist Stephen Hawking, scientific and technological progress is so rapid that computers will ‘surpass humans’ at some point within the next 100 years. Do you think it’s not soon? Nice one. Individual elements of AI are already ready to take control of the lives of hundreds of millions of inhabitants of our planet. At the same time, not everyone will be able to recognise where he or she makes a decision, and where it is done by artificial intelligence. And this is often fraught with unpredictable consequences.

Invisible assistants 

Artificial intelligence can mimic human behaviour to perform tasks and learn incrementally from the information it collects. And now the music streaming services offer compositions for listening, tracking the preferences of the owner of the gadget. Social networks recommend new friends, bring together potential employees and employers. Advanced cars automatically adjust various settings based on the likes and dislikes of their owners. The navigator in the car calculates the best path to your destination. Email helpfully sorts emails by subject, calculates spam, and optionally includes a one-click smart reply feature. Search engines work taking into account the data previously collected on you — location, age, previous requests, etc. Online stores are ready to give recommendations on the choice of goods after typing the first letters of the desired name…
If all these virtual assistants with the habits of a warden are annoying, then not always, not all of them, and not as much as it might seem. The talk about the power of neural networks is much more disturbing. (By the way, there is an important difference between them and artificial intelligence. A neural network is just one of the areas of development of artificial intelligence, the purpose of which is to model the analytical mechanisms carried out by the human brain.) Among the tasks that a typical neural network solves are classification, prediction, and recognition. The main areas of their application are forecasting, decision-making, pattern recognition, optimisation and data analysis. All the things we love. But what a range of applications are there! From speech recognition systems to protein secondary structure recognition, from the classification of various types of cancer to genetic engineering. By the way, now popular chat bots are also the work of neural networks.

Every year, the scope of neural networks is increasing. According to a report by Allied Analytics, the market for neural networks will be $39 billion in 2023, nearly six times the size of 2016.

Put your trust in AI, and keep your powder dry 

Any technology can both benefit and harm — it all depends on how you use it. Internet users, and not only them, faced one of the global problems. It’s about fakes. The appearance of, say, programmes for replacing faces and even the season in photos and videos gives a wide scope for imagination. Generative neural networks already allow you to create videos in which people do things that they have not done and say things that they have never said. 

Hence the conclusion: it is impossible to trust the AI news feed today. You can only trust reliable, well-established sources.

It is believed that neural networks are a good tool for hackers: professionals successfully overcome anti-virus protection systems and create new generations of malware.
But still, the main horror story for people who are associated with the rapid development of AI in general and neural networks in particular is the upcoming global changes in the labour market. Indeed, the further introduction of artificial intelligence technology will increase productivity — primarily by getting rid of people whose activities can be automated as much as possible. According to experts from Goldman Sachs (USA), 18 percent of the current work in the world can be automated in the future with the help of artificial intelligence. 

Lots of noisy data… For what reason?

Artificial intelligence cannot create, conceptualise, or strategically plan. It’s great at optimising for a narrow goal, but it’s incapable of thinking creatively. So it is still far from a complete replacement of a person. Nevertheless, in March of this year, the American entrepreneur, engineer and billionaire Elon Musk, together with more than 1,000 AI experts, demanded a moratorium on training powerful neural networks in an open letter, since an uncontrolled process in this area could pose a threat to humanity.
While celebrities in the West are attracting attention with loud statements, practitioners are pursuing other tasks. As Director of the Hi-Tech Park Vsevolod Yanchevsky said last autumn at the opening of the first exhibition-forum of IT Academgrad ‘Artificial Intelligence in Belarus’, a touch of sensationalism and partly scandal around AI is passing. It’s time for real results. 


In turn, Director General of the Informatics Institute of the National Academy of Sciences of Belarus Sergei Kruglikov recalled that the domestic AI community occupies a leading position in the international arena: “If we take smart healthcare, intelligent image processing with high accuracy, then NAS employees are in the top 10 best manufacturers. Our brands are known and used all over the world. In terms of intellectual healthcare, we are the first among the CIS countries,” he gave an example. 
More than a hundred projects in the field of artificial intelligence and robotics were presented at the autumn exhibition. At the beginning of 2023, the development of AI was already discussed during the Intellectual Belarus exhibition. For many uninitiated, it was news that, for example, all healthcare services in Minsk and other cities — Automated information and analytical systems such as Clinic and Web-polyclinic, Ambulance, Unified Laboratory Information System, decision support systems for predicting one or another disease are the results of the work of domestic specialists. And ‘smart’ cities, healthcare, education, housing and communal services, etc. are realistic projects. As well as speech and text recognition and synthesis technologies, other developments in the field of AI.

The Head of State delegated the functions of an operator-regulator in the field of intellectual technologies to the Informatics Institute of the Natioal Academy of Sciences at the exhibition Intellectual Belarus. 

There is a lot of work to be done in the near future, Sergei Kruglikov notes in this regard, “It is necessary to prepare a regulatory framework, develop standards and an action plan for the introduction of intelligent systems in the daily activities of our people.”

Neural networks can work according to a given algorithm, remember information, and also generate reactions, recognise errors and learn from them on their own.

By Aleksandr Nesterov