Behind every artificial intelligence, no matter how advanced, is the originality of the human mind. The AI Journey conference serves as a microcosm of our society, bringing together individuals from all walks of life who engage with AI in various aspects of their work.
Each year, the event draws developers and researchers from across the globe, and the 2024 conference, organised by Sberbank in Moscow last week, was no different. It featured presentations on innovative solutions and tools designed to benefit IT specialists. AI Journey serves as a one-stop platform for exploring global trends in AI and IT while also providing industry professionals with an opportunity to discuss the most pressing issues and challenges facing the international community and the tech industry.
Developers are the backbone of AI innovation. Russian developers now have access to GigaView, a new tool based on the GigaCode AI assistant designed to enhance the efficiency of teamwork on software projects. The solution was introduced by Andrey Belevtsev, senior vice president and head of technology development at Sberbank.
“In today’s world, it’s extremely important to provide our developers with effective tools for boosting growth in productivity and to give businesses the opportunity to efficiently apply these solutions for process optimisation. GigaView is aimed at automating the most labour-intensive and routine operations and improving the management of the development process. The tool reduces coding errors and makes creating code transparent for all team members,” Belevtsev explained.
GigaView is designed to detect potential defects during development and evaluate the quality and productivity of AI projects. Notably, it allows productivity tracking at both the team and individual levels using detailed metrics. The tool is now available on GitVerse, a Russian platform for managing source code. Currently, it supports Java, JavaScript, and TypeScript, with plans to expand to additional programming languages soon.
Belevtsev also highlighted updates to existing tools, including an upcoming upgrade to Sber’s GigaIDE Desktop. This enhanced version of the Russian integrated development environment will enable users to tackle Java project code tasks more efficiently with the help of AI agents, significantly reducing development time. IT professionals will find it easier to locate the project data they need. GigaIDE Desktop has already demonstrated its safety and efficiency during testing by Russian government agencies and has been added to a special register.
These new tools might seem exclusive to those with specialised training, but that’s not entirely the case. The AI Journey conference highlighted solutions for professionals across various fields, including the creative industries. One standout example is Kandinsky 4.0 Video, a neural network designed to deliver innovative creative solutions. Sber unveiled the beta version of this next-generation tool that allows users to create realistic video clips from text prompts or a starting frame.
With its user-friendly interface, Kandinsky caters to everyone from casual users making amateur videos to designers, videographers, and film animators. It can produce anything from polished video clips to professional-grade project trailers.
In just a year since its debut at the previous AI Journey conference, Kandinsky has made significant advancements. Developers have improved the quality of its video output while dramatically increasing generation speed. The model can now transform a user’s idea into a 12-second HD video clip at 1280×720 resolution. For those seeking even faster results, Kandinsky 4.0 Video Flash offers an expedited option capable of creating a video clip in just 15 seconds. This rapid version of the neural network will be available to the public in 2025.
“These models, such as Kandinsky Video, are facilitating global development in this important field, taking modern technologies much closer to the synergetic level of information processing, perception and creation possessed by human beings,” Belevtsev further said.
Developers believe that many AI technologies should remain open-source as this approach will accelerate the growth of the AI community. A prime example of this philosophy is Sber’s commitment to publishing its artificial intelligence models in open source. The code and weights for several models, including the basic and instructional versions of the fastest model, GigaChat Lite, as well as GigaEmbeddings-instruct – ranked as the top Russian-language model by the ruMTEB benchmark – are now accessible to all users.
Sberbank’s chairman, Herman Gref, announced the use of AI in public administration. For the first time, Russia will employ Sber’s large language model, GigaChat, to process questions submitted by Russian citizens for a live broadcast with their President Vladimir Putin.
A highlight of the conference was the introduction of an advanced algorithm for accelerating transport routing. In collaboration with ITMO University, Sber and its AI Laboratory specialists developed the RIDE Python library, which enables faster creation of optimal routes. What sets this solution apart is its ability to improve logistical calculations at scales ranging from city-level to nationwide. The library uses an innovative clustering approach to divide the graph and give developers enhanced control over the quality of their work. This method has been successfully tested in over 500 cities.
Shopping on online marketplaces can be a time-consuming task due to the sheer volume of products available. However, developers have created a way to help users quickly navigate through countless items and find the best deals. At the AI Journey conference, researchers from Sberbank’s AI Laboratory introduced new recommendation systems. These systems were trained using open data sets to improve neural networks that assist users in product discovery.
The event wasn’t just about current developments; it also laid the groundwork for future collaborations. A major announcement revealed plans for a new Artificial Intelligence Alliance, uniting participants from 18 countries, including the BRICS nations. This initiative seeks to promote the development of AI technologies openly and transparently for everyone’s benefit.
Events like AI Journey aim to drive responsible, innovative, and inclusive AI development and tackle some of the world’s biggest challenges by uniting diverse perspectives and encouraging international cooperation.
Mastering Diffusion Models: Prompting & Fine-Tuning Techniques | AIJ 2024 (0+)
The post LLMs, AI Innovation Take Centre Stage at AI Journey 2024 appeared first on Analytics India Magazine.
