The new department focuses on industrial design using artificial intelligence and training in prompt engineering for neural networks.
The Russian laboratory for industrial design, engineering, and inventions, MNTlab, has announced the launch of Russia's first IDGPT (Industrial Design based on GPT technologies) department. The new division applies modern artificial intelligence technologies in the field of industrial design, creating a productive combination of human creativity and the potential of neural networks.
The decision was made following a study conducted with hh.ru, which revealed a shortage of personnel and a trend towards a sharp increase in salaries, which small companies sometimes cannot afford.
The department has four people. It includes current specialists who have undergone retraining.
MNTlab specialists delegate part of the basic project tasks to the neural network—primarily the initial design of sketches and the creation of demonstration materials. Design and prototyping are the most energy-consuming part of the development, when it is necessary to process a huge amount of incoming requests from the customer's TOR and turn it into a short list of key ideas/sketches of the product, which then human industrial designers take into development and manually lead to implementation. Collaboration between humans and artificial intelligence leads to innovative ideas and unique solutions that previously seemed inaccessible or associated with high overhead costs.
The new approach allows to increase productivity and increase production capacity, while maintaining key creative competencies for people, which form the company's competitive advantage in the market, emphasizes Mikhail Krapivnoy, CEO and co-founder of MNTlab. Without AI, the designer performs the initial creation of sketches in 10-15 hours. With AI—in 2-3 hours. The typical project is delivered in an average of 8-10 weeks (this is a very average figure, there are more complex projects), and with AI there is a reduction of at least 2-3 weeks, which is extremely important for customers.
Artificial intelligence is actively involved in design and prototyping, creating the basis for product sketches, which are then refined by industrial designers. Thanks to the new department, the company expects to reduce operating expenses by about 25% by the end of the year.
It is expected that the creation of the IDGPT department will lead to an increase in the company's throughput by a quarter by 2024. In addition, MNTlab plans to look for new optimization points, which will reduce costs by 15-20% within a year after the full implementation of new artificial intelligence methods.
A corporate training center has also been created on the basis of the department, which, according to Krapivnoy, will later become available to designers and managers from other companies in the domestic market. The center was created to train employees in methods of managing search queries, prompts (promt), so that designers and engineers can form the most effective search queries for performing tasks by artificial intelligence and get an adequate visual or textual basis for subsequent revisions by a person. "In simple words, we teach customers to formulate their wishes for the product being developed and the meanings that the brand wants to convey to its consumer in a concise and understandable way," says Krapivnoy.
Despite the fact that modern language models of neural networks have the ability to process long contexts, it is still unknown how effectively they use this opportunity. Mikhail Krapivnoy recalls that during international studies of language models of neural networks in 2023, it turned out that the performance of models is often highest when relevant information is at the beginning or end of a long search query. However, the probability of successful extraction of relevant information is significantly reduced when models need to refer to the middle of long search phrases. Moreover, as the length of the input context increases, even models with a pronounced ability to work with long texts experience a significant deterioration in performance.
"On the one hand, long, detailed requests help to get the highest quality solution to the problem from the neural network, on the other hand, the longer the request, the more often the machine has errors, which is why the ability to competently design multi-level prompts becomes the most important soft skill for industrial designers," notes Milena Manukyan, leading industrial designer at MNTlab.
Artificial intelligence has become a key technology that determines the future of industry and business. European technology giants are aware of its potential and strive to take leading positions in this area. Western companies invest their funds in artificial intelligence research and development. They attract the best specialists, create their own research laboratories and actively train their employees to interact with neural networks in order to enrich their intellectual capital for creating innovative products and services. An example of such a transition to the use of artificial intelligence in creative industries is the American online publication BuzzFeed, which back in January of this year decided to cut about 180 journalists and redirect efforts to work with the GPT chat. This step demonstrates that companies are actively moving from traditional workflows to the use of modern technologies to improve efficiency and improve their product or service.
A similar process has been reflected, for example, in the electronics market. Foxconn, Apple's largest manufacturing partner, has decided to replace 60,000 employees with robots. This is a desire to automate production processes and use modern technologies that can improve the efficiency and quality of production. "Of course, our company does not give and does not plan to give all work tasks completely under the control of the machine. The department functions in conjunction with the "human-machine", but we can already say that this solution allows our team to increase productivity and increase our production capacity," Krapivnoy summarizes.
In June of this year, during the plenary session of the ASI forum "Strong Ideas for a New Time", Russian President Vladimir Putin stated that industrial and product design should become an important resource in the development of domestic business in the coming years.
Industrial design is becoming an integral resource for business, contributing to its survival and restructuring of its business processes. It is an effective tool for the successful operation of enterprises in the new reality, and the introduction of technologies and processes is becoming increasingly relevant.
