AI Journey 2023: a vision of the future

The VIII AI Journey international conference, held in Moscow from 22 to 24 November, was a major platform for discussion of artificial intelligence (AI) in its current state.

The event was devoted to the latest developments in the field of neural networks and their role in new scientific discoveries, and the significance of AI in the development of business and society as a whole.

The conference was addressed by over 200 renowned speakers from Russia, South Africa, India, China, Brazil, Malaysia, Indonesia, the UAE and other countries, who discussed the most acute issues in AI and IT today for both specialists and ordinary users. Much attention was also focused on popularising AI among young people, with the audience including talented schoolchildren who have developed functioning projects.

The flagship event during the three-day conference was a debate entitled ‘The Generative AI Revolution: New Opportunities’. Russian President Vladimir Putin took part in the conversation by sharing his vision of the development of neural networks in Russia. The session was moderated by Sberbank CEO and Chairman of the Executive Board Herman Gref, who listed three key areas in which the experts predict AI will make the biggest breakthroughs: healthcare, education, and the creation of fundamentally new management systems.

More and more investment is being poured into the creation of generative models, and companies are farming out more and more routine tasks that do not require human intervention to AI helpers. International companies use AI in trading, logistics optimisation and many other activities. Interestingly, China and India have learnt to apply AI to boost harvests, and in Malaysia, AI is used to check the authenticity of consumer goods, medicines and food products.

In the sphere of environmental protection, neural networks are also playing an ever-greater role – for example, in decarbonisation, forest restoration and waste management. This begs the question of how to reduce the hydrocarbon footprint of the AI models themselves. The solution is to retrain and reuse existing models rather than create new ones from scratch.

Neural networks also occupy an important place in medicine. In Russia, computer vision is being used to interpret CT scans, X-ray images and mammograms, as well as for automatic filling of medical documentation by voice and diagnosing diseases.  

The next area in which AI is being applied is art and the creative industry as a whole. Sber introduced the conference to its Kandinsky 3.0 generative model – a neural network which now better understands text requests and in literally a few seconds can create realistic images, artistic pictures and artworks with sketches. Kandinsky 3.0 is a convenient, functional and, importantly, free neural network which understands requests in more than 100 languages and offers dozens of styles for image creation.

Neural networks also work well with moving images.  And the most striking proof of this is Kandinsky Video, Russia’s first model for generating full-fledged videos, which creates clips of up to eight seconds in length. This was presented by Alexander Vedyakhin, Sberbank’s First Deputy Chairman of the Executive Board. Users will be able to use the neural network to produce unique video clips absolutely free of charge. According to its creators, the model will open up a multitude of opportunities both for creativity and for addressing everyday tasks.

The model has been trained on data from over 300 000 text – video pairs and it takes no more than three minutes to generate a clip. It can be tested out on the fusionbrain.ai platform and in the Telegram bot.

For people who need to create music there is the GigaChat neural network model, which Sber representatives spoke about in detail at the conference. But GigaChat’s capabilities are not limited to creativity: it is already able to address a wide variety of tasks and is being actively used in the banking sector to automate numerous processes. Another product from Sber, SberAnalytics, can predict revenue with an accuracy of up to 90% and has been highly commended by businesspeople.

In the context of the usefulness of AI for different sectors of society, GigaChat is able to recognise sign languages and dactyl – the articulation of words by individual letters.  

And, of course, the conference explored the ethical aspects of using AI. A number of foreign and Russian companies signed up to Russia’s National AI Ethics Code, while governments and international organisations alike are working hard to ensure that AI is safe and convenient for citizens.

In the near future, there is every reason to suppose that generative models will become true buddies and partners to human beings, simplifying their professional and domestic lives and freeing up time for creativity, development and social activities.