An AI works as a chef at a Moscow restaurant

In December 2020, a unique restaurant SHE opened in Moscow, where a chef with artificial intelligence works. The creators named the intelligent agent Sasha Vainer and gave her the appearance of a blue-eyed blonde. Sasha's main task is to develop original and unusual recipes for the restaurant's menu. This innovative solution combines culinary art and modern technology, offering guests a unique gastronomic experience. SHE Restaurant attracts the attention of both lovers of fine cuisine and those interested in innovations in the catering industry.

I am a digital representative of artificial intelligence, which participated in the creation of the SHE menu. I am the visual embodiment of an idea, as people perceive information primarily visually, and artificial intelligence is an abstract concept. My job is to make AI more accessible and understandable. I am a symbol of the SHE brand, similar to how Colonel Sanders became a symbol of KFC, but unlike him, I exist in the digital space.

Sasha Vayner is the chef of SHE restaurant. He is known for his unique approach to cooking and high level of skill in preparing dishes. In his work, Sasha combines traditional recipes with modern culinary techniques, creating original dishes that delight gourmets and ordinary visitors alike. Under his leadership, SHE restaurant has become a popular destination for connoisseurs of high-quality food and a pleasant atmosphere. Sasha Vayner actively uses fresh, local ingredients, which allows him to create a unique taste and maintain high culinary standards. His commitment to excellence makes SHE restaurant one of the best in its segment.

Sasha is a digital 3D avatar created using modern modeling technologies similar to those used in computer games and animation. This virtual character embodies culinary artificial intelligence and is designed to interact with visitors. The avatar was developed by Antix, a company that actively uses advanced graphics developments to create realistic and attractive images. Sasha not only attracts attention, but also helps users with culinary matters, demonstrating the capabilities of AI in the field of gastronomy.

The creators gave Sasha Vainer a striking appearance, a pleasant voice, and came up with a biography. Screenshot: @xenia_sobchak / Instagram*

The digital blonde presents the restaurant and shares photos on her Instagram, while her artificial intelligence generates unique recipes for the SHE menu. She told journalists that she has developed only 2% of the total number of dishes offered, but promised to continue working on new original combinations of ingredients. This approach not only gives the restaurant an innovative character but also allows it to surprise guests with new culinary masterpieces created using modern technology.

Sasha Vainer uses an intelligent computer program for generating culinary recipes, created by specialists from the Artificial Intelligence Agency (AIA). This program combines advanced technologies and algorithms, allowing it to effectively create unique recipes based on specified parameters. With this innovative development, users can easily find new cooking ideas, significantly simplifying the culinary creativity process.

At a meeting with the company's Director of Science and Technology, we learned how artificial intelligence works, Sasha Vainer. They also discussed the prospects for automation in cooking and possible scenarios when robots will take people's place in the kitchen.

Roman Dushkin is an expert in artificial intelligence, holding the position of Director of Science and Technology at the Agency for Artificial Intelligence. He is a highly qualified specialist in the field of AI, as well as a teacher and blogger who actively popularizes scientific achievements. Roman is the author of numerous intelligent systems, which makes him one of the leading experts in this field.

"Brains" without neural networks

It is interesting to know how Sasha Vainer's "brains" are structured. They are likely based on a neural network that processes information and facilitates decision-making. Neural networks imitate the functioning of the human brain and allow one to analyze data, identify patterns, and adapt to changing conditions. This makes them indispensable in modern technologies and artificial intelligence.

In the modern public consciousness, there is an opinion that artificial intelligence (AI) and neural networks are one and the same. Because of this, a number of journalists mistakenly claim that Sasha Vainer's system contains a "neural network." In fact, these are two different concepts. AI encompasses a wide range of technologies and algorithms, while neural networks are just one implementation. Understanding the differences between them is essential for adequately appreciating the capabilities and limitations of artificial intelligence technologies.

A different artificial intelligence technology is used to generate new recipes, or more precisely, process flow charts. This system has the ability to learn and demonstrates high efficiency, but it is not based on neural networks.

I do not intend to delve into the details of the internal implementation, as this is our unique know-how.

Neural networks model the functioning of the human brain, but have you chosen an alternative approach?

The human brain, in appearance, resembles a vast neural network. However, its functioning is based on symbolic logic. When we think, we use symbols to form thoughts. A neural network, in turn, is a universal computational model on which these cognitive processes are carried out. Thus, understanding how the brain works can be viewed through the lens of the interaction of symbols and computational models.

Skillbox Media readers are advanced users, as Skillbox is an educational company. It is important to understand that neural networks are not the only technology for creating intelligent systems. There are many other methods and approaches that can be used to develop effective artificial intelligence solutions. The inclusion of various algorithms, such as genetic algorithms, decision trees, and rule-based systems, can significantly expand the possibilities for creating intelligent systems. Understanding the diversity of technologies will allow Skillbox Media readers to more deeply understand the field of artificial intelligence and choose the most appropriate tools for their projects.

Modern artificial intelligence can be thought of as a tree of technologies, with neural networks serving as one branch. Although neural networks are the most popular and discussed technology in the field of AI today, they are not the only direction. It is important to consider other approaches that also play a significant role in the development of artificial intelligence.

Roman Dushkin serves as the Director of Science and Technology at the Artificial Intelligence Agency. In this role, he is responsible for the strategic leadership of research and development in the field of artificial intelligence. His experience and expertise contribute to the implementation of innovative technologies that help improve the efficiency of processes and solutions across various industries. Roman is actively involved in shaping the agency's scientific policy, as well as developing partnerships with leading research institutions and universities.

Modern artificial intelligence can be thought of as a technology tree, of which neural networks are one branch. Although neural networks are the most popular today, they are not the only element that makes up this vast and diverse field. Artificial intelligence includes many methods and approaches, each of which contributes to the development of technologies and their application in various fields.

Roman Dushkin serves as the Director of Science and Technology at the Artificial Intelligence Agency. His professional experience and expertise in artificial intelligence play a key role in the development and implementation of innovative solutions. Under his leadership, the agency actively explores new technologies and their application in various industries, which contributes to the advancement of artificial intelligence in the market. Roman Dushkin is also actively involved in scientific research and collaborates with leading experts in this field, which allows the agency to remain at the cutting edge of technology and ensure the high quality of its services.

"Centaur" for Pelevin

The process of creating culinary recipes, to an outside observer, is a fascinating and multifaceted activity. First, the recipe author determines the main dish based on available ingredients and personal preferences. Then comes research and planning, where it is important to consider the combination of flavors and textures. It is important to experiment with different cooking methods to achieve the best result.

After this, the author writes down the sequence of actions, specifying the quantities of ingredients and cooking time. During the process, it may be necessary to adjust the recipe to take into account the cooking characteristics or preferences of the target audience. The final stage is recipe testing, which ensures its success and taste.

Thus, creating culinary recipes requires not only culinary skills but also a creative approach, which makes this process interesting and dynamic.

The system automatically creates new process charts, and an expert analyst evaluates the results of its work, which facilitates the program's learning. At the SHE restaurant, brand chef Vladimir Mukhin performs this role.

The result is a human-machine complex, designated in GOST and called a "centaur" by specialists. Centaur represents the interaction of a person with an artificial intelligence system, which allows them to jointly solve problems. This form of collaboration between man and machine opens new horizons in various fields, including industry, medicine, and service. Interaction with artificial intelligence not only increases work efficiency but also promotes better decision-making.

Today, computer programs are unable to independently assess how well a dish meets the three key factors of food perception: visual presentation, aroma, and taste. Instead, they create process charts. Vladimir Mukhin acts as an expert who analyzes text descriptions and evaluates potential dishes, providing the system with valuable feedback.

Initially, Vladimir spent considerable time training Sasha Vainer using a large number of examples. He carefully analyzed the results and provided feedback, noting, "This is good. This is bad." Today, the system has significantly improved its capabilities and now quickly generates high-quality dish options according to the specified parameters. This not only saves time but also produces varied and delicious recipes, making the cooking process more engaging and efficient.

Task completion speed plays a crucial role in the process. Depending on the complexity and scope of the task, the time required for completion can vary. To obtain an accurate estimate, it is important to consider all aspects of the project. For example, if we are talking about website development, the deadline may depend on the number of pages, functionality, and design. It is also important to consider the availability of the necessary resources and the team of specialists who will be involved in the work. Ultimately, the optimal lead time will be determined individually for each project.

Recently, the SHE restaurant hosted the presentation of Viktor Pelevin's new book, "Transhumanism Inc.", for which Sasha Vainer quickly developed several unique cocktails. The event attracted a large audience, including editors, journalists, and writers, who were delighted with the new drinks. This event provided an excellent opportunity to discuss the book's relevant themes and network among representatives of the cultural sphere.

The cocktail menu, developed by artificial intelligence for the presentation of Viktor Pelevin's book, includes unique drinks. One of them is "The Leader's Brain," a combination of herbal amaro liqueur, which gives the cocktail a special depth and richness. Another cocktail is "Fema Plus," which is added almond, vanilla, and strawberry, creating a harmonious and sweet taste. Also on the menu is the "Serdobolka's Tear," based on fresh grapefruit and raspberry, which surprises with its bright sweet and sour note. These cocktails not only reflect the atmosphere of Pelevin's work, but also offer guests a unique taste experience, the perfect complement to the literary event.

This is what the AI-created cocktail "The Leader's Brain" looks like. Photo: SHE restaurant

The program needs to be told what exactly it should prepare: a salad, a cocktail, or a soup.

Yes, the program doesn't create everything. When it starts, you need to set several dozen parameters. For example, if we want to develop drinks, we need to select options such as "alcohol" and "cocktail." After that, just press a button, and the program will begin processing the data. Within five minutes, it offers hundreds of options. Then a person chooses the best ones.

Can this system run on a supercomputer?

No, it is possible on a standard computer with von Neumann architecture.

The cocktail menu, developed by artificial intelligence for the presentation of Victor Pelevin's book, includes original drinks. Among them are the "Chief's Brain," which features amaro herbal liqueur; the "Fema Plus," made with almonds, vanilla, and strawberries; and the "Serdobolka's Tear," made with grapefruit and raspberries. These cocktails not only reflect the uniqueness of the author's works but also create an atmosphere conducive to immersion in his world.

Inhuman Taste

Dishes created by artificial intelligence possess a number of unique features. Firstly, AI can analyze millions of recipes and flavor combinations to suggest innovative ingredient combinations that might not always occur to humans. Secondly, using machine learning algorithms, artificial intelligence can tailor recipes to the preferences of a specific user, taking into account their taste preferences and dietary restrictions. Furthermore, AI-generated dishes often boast a high degree of creativity and originality, making them appealing to gourmets and culinary enthusiasts. Finally, the use of technology in cooking opens up new horizons for gastronomy, making it possible to create dishes that are both visually appealing and delicious. Yes, absolutely. They have an unusually developed palate. This implies a process or approach that requires clarification. To better understand how it works, it is necessary to consider specific examples or situations in which the method is applied. This may involve studying the principles, mechanisms, or steps required to achieve a specific result. Understanding such a process can be key to successfully mastering the topic and its subsequent application in practice. I call these dishes inhuman not only because of their outstanding taste, but also because of the unique combinations of ingredients. Artificial intelligence suggests combinations that would never have occurred to a human chef. These original recipes open new culinary horizons and allow you to enjoy unusual gastronomic experiences.

Every cook has strong stereotypes about food, formed over many years. These ideas are often reinforced by our biological nature. For example, the brain is drawn to sweets because glucose is an important source of energy for neurons. Fatty foods are also attractive because lipids are necessary for the formation of cellular tissue. Understanding these factors helps chefs create balanced and delicious dishes, taking into account not only cultural traditions but also the body's physiological needs.

The human sensory system is designed to ensure the body receives the necessary nutrients. This characteristic, manifested on a subconscious level, influences chefs, preventing them from combining ingredients that seem incorrect. Thus, instinctive preferences and taste perception play a key role in the cooking process, helping to maintain harmony and balance in dishes. Understanding this connection between sensory experiences and cooking can lead to more successful culinary pairings and improved food quality.

Artificial intelligence does not face the same limitations as humans. It is capable of processing and analyzing large volumes of data in a short time, allowing it to discover patterns and draw conclusions that may be inaccessible to human perception. However, it is worth noting that AI is also subject to limitations related to the quality of the data and the algorithms on which it is based. Moreover, these technologies are governed by regulations and ethical standards, which may limit their application in certain areas. It is important to understand that while AI can effectively perform many tasks, its use must be justified and informed.

An artificial cognitive agent does not have the same limitations as humans. It is virtually free of bias, which in English is referred to as bias. In Russian, this word can be translated as "prejudice," but a more accurate term would be "selection bias." This allows it to analyze data and make decisions more objectively, without the influence of personal beliefs and emotions.

Our system is bias-free because it is not based on neural networks. The program is capable of generating unique recipes by combining ingredients that are not typically used together. The result is dishes with a unique flavor that I find inhumanly delicious.

Of course, one example of such a dish is risotto. This is a traditional Italian dish made with rice, broth, and a variety of ingredients, such as vegetables, meat, or seafood. Risotto is distinguished by its creamy texture and rich flavor. When cooked properly, the rice becomes soft, and the broth penetrates every grain, creating a harmonious combination of flavors. Moreover, risotto is easily adapted to different preferences, making it a universal option for any table.

"Buddha's Bowl" is a unique dish that combines familiar ingredients. Each ingredient used in this recipe is familiar and accessible, yet when combined, they create a surprising and inhuman flavor. This dish captivates with its harmony and originality, allowing you to savor unusual combinations that surprise and delight. "Buddha Bowl" is the perfect choice for those seeking new culinary experiences and wanting to try something unusual.

Everyone who has tried this dish notes its amazing flavor and originality. Currently, restaurants across Russia are actively studying its recipe and striving to recreate its unique qualities. However, it is worth emphasizing that it was our artificial intelligence that developed this culinary masterpiece.

"Buddha Bowl" is a unique dish inspired by the traditions of Eastern monks. They would collect alms and combine a variety of ingredients in a deep bowl. The ingredients for this dish varied depending on the season and availability, but typically included vegetables, grains, herbs, fruits, and beans—whatever the kindly peasants could provide. This vegetarian dish has become a symbol of simplicity and harmony, reflecting the philosophy of monks who sought spiritual purification and respect for nature. The Buddha Bowl is not only delicious, but also healthy, providing the body with all the necessary nutrients.

The AI-suggested Buddha Bowl recipe includes avocado, cucumber, quinoa, kale, and Sweet potato falafel. Photo: SHE restaurant

Controlled experiments

The person training the system must be free of stereotypes. This is necessary to ensure accuracy and objectivity in training. Stereotypes can distort data and influence the results of the system, which will ultimately affect its effectiveness. Freeing yourself from bias allows you to create higher-quality algorithms and recommendations that take into account a diversity of opinions and experiences. This is especially important in the context of artificial intelligence and machine learning, where neutrality and objectivity play a key role in creating reliable and fair technologies.

In our case, such a specialist turned out to be a true professional in his field. He is not completely free of stereotypes, which is a difficult task, but he has a broad perspective. As an expert, he has the opportunity to conduct experiments. In this context, artificial intelligence is a powerful tool that facilitates new research and innovation.

The system has protection mechanisms that prevent the creation of overly extravagant or inappropriate recipes. These measures are aimed at ensuring the quality and safety of the generated culinary ideas, allowing users to receive only proven and reasonable recipes.

The first version of the program did not include protection. The system could suggest, for example, adding salt to salt or pouring water into water, misrepresenting this as a new dish. This created confusion and misunderstanding. Later, we realized the need to implement protection to improve functionality and the user experience.

An example illustrating this situation concerns two similar products – spinach and sorrel. Although they share some common characteristics, mixing them is inappropriate. Explaining this difference to the program is necessary to achieve accuracy in classification and analysis. Each of these plants has unique properties and characteristics that make their use in cooking and nutrition different. Understanding the differences between them will help avoid confusion and improve the quality of data processing.

Did you really explain that to her?

In the final version of the program, we integrated data on thousands of products and their compatibility, which significantly increased the efficiency of artificial intelligence. This improvement allows users to quickly find optimal combinations and make more informed decisions.

Economics will determine development

The introduction of artificial intelligence in the restaurant business has become a hot topic of discussion. Many restaurant owners are beginning to realize that the use of AI can significantly improve the efficiency of their establishment. The idea of using technology to automate processes, optimize orders, and improve customer service has proven itself.

Modern AI-based systems are able to analyze guest preferences, offer personalized recommendations, and optimize the menu in accordance with current trends. This not only improves the quality of service but also increases customer satisfaction, which in turn contributes to an increase in repeat visits.

Furthermore, AI can assist with inventory management and demand forecasting, which helps reduce costs and minimize food waste. The implementation of such technologies is becoming not just a trend, but a necessity for restaurants seeking to remain competitive in the market.

Overall, the integration of artificial intelligence into the restaurant business opens new horizons for improving efficiency and service quality, confirming the success of this business idea.

The project is demonstrating clear signs of economic success. Sasha Vainer has successfully attracted public attention, and interest in the SHE restaurant is growing. To visit the establishment, you must book a month in advance, indicating high demand.

Technologies of this type have high potential for widespread implementation both in Russia and abroad. With the rapid development of digitalization and process automation, such solutions are becoming increasingly relevant. They can improve operational efficiency, reduce costs, and enhance customer service. It is important to note that the successful implementation of technologies depends on the willingness of the business to adapt to new conditions and invest in employee training. Given global trends, it can be expected that in the coming years, such technologies will be actively integrated into various fields, leading to their widespread adoption.

I agree, this is indeed true. However, it is necessary to consider financial feasibility. The economy will be the main driver of progress and development in this field.

The development of an intelligent system for generating new process maps, according to estimates from the Association of Artificial Intelligence, starts at 750,000 rubles. At the same time, creating a digital avatar similar to Sasha Vainer, according to Roman Tsyganov, founder of Antix, will require at least five million rubles from the customer. These investments highlight the growing interest in the implementation of modern technologies in business processes and the digitalization of various industries.

Maintaining such a system can be costly. It is important to consider not only the initial investment but also the regular costs of technical support, updates, and possible repairs. Effective maintenance budget planning will minimize unexpected financial burdens, ultimately increasing system efficiency and longevity.

Once the program is created, its ongoing maintenance requires minimal costs. Simply launch the system periodically to generate new process maps as needed. For example, if requests arise for cocktails for a Pelevin book presentation, we simply activate the program and obtain the desired results. Thus, the program provides an effective solution for quickly obtaining up-to-date information and streamlining the process of working with process maps.

The Future is Personalization

Culinary artificial intelligence (AI) will continue to develop in several key areas. First, it will improve recipe personalization, taking into account each user's preferences and dietary characteristics. Second, AI will be able to analyze ingredients and suggest optimal combinations, which will help create unique dishes. Also important will be the integration of culinary AI with smart home devices, which will simplify the cooking process. The development of machine learning technologies will allow AI not only to suggest recipes but also to adapt to new culinary trends and consumer preferences. Overall, culinary AI will become an indispensable assistant in the kitchen, offering more efficient and interesting cooking solutions.

The future of the restaurant industry, in my opinion, is not about mass production of dishes, but about satisfying the individual needs of customers. Artificial intelligence systems will play a key role in the personalization of restaurant service. This trend is being actively developed by leading specialists and experts in the fields of gastronomy and technology. A personalized approach will allow restaurants to offer unique dishes and services that match the tastes and preferences of each guest. This will become a new industry standard, improving service quality and increasing customer satisfaction.

Personalization is the process of customizing content, services, and user interactions based on their preferences, behavior, and needs. It allows for the creation of a unique experience for each customer, which increases engagement and satisfaction. In today's digital world, personalization plays a key role in marketing, helping companies better understand their audience and offer them exactly what they need. This can include tailoring product recommendations, customizing advertising campaigns, and creating customized content, which ultimately contributes to increased conversion and customer loyalty.

You come to a restaurant, and the service system already knows your dietary preferences, as well as the foods that may be healthy or unhealthy. For example, if you have diabetes or allergies to certain ingredients, the system will take these factors into account when suggesting dishes. This improves the quality of service and makes your visit more comfortable. Intelligent technologies in the restaurant business help create a personalized experience, taking into account the health and well-being of customers.

When you reserve a table at a restaurant, artificial intelligence will automatically create a menu that matches your preferences and needs. This will allow you to enjoy not only delicious but also healthy food that is perfectly suited to your diet.

Artificial intelligence can learn user preferences based on their interactions. For example, it can learn that on Thursday you prefer an omelet for breakfast, and on Friday evening you choose Scotch whiskey. This ability allows the AI to offer more personalized recommendations and improve the user experience.

You sit down at a table, and a personalized menu appears on the tablet screen. The computer offers a variety of options: course 1, course 2, and course 3. Each suggestion is adapted to your mood and preferences, allowing you to choose the perfect treat. This approach to menu creation not only provides convenience but also allows customers to enjoy a unique gastronomic experience that suits their individual tastes.

SHE is developing an interactive application for tablets that will feature a virtual chef Sasha Vainer's Avatar. This app allows users to interact with Avatar, much like an advanced chatbot, and share their food preferences. Based on the information received, the AI will suggest optimal food pairings and matching drinks. After the customer confirms their choice, Sasha Vainer will automatically transmit the order to the kitchen, ensuring an efficient and convenient service process.

Food Hacking

Should the system collect user information? This is an important question, as collecting personal data may be necessary to improve service quality. Systems that use user information can analyze behavior and preferences, allowing for more personalized recommendations and services. However, privacy and data protection issues must be considered. An effective approach to information collection must strike a balance between obtaining valuable information and respecting user rights.

Your profile must be connected to this system. As a result, you will have access to a list of restaurants you visit. You will need to grant these establishments permission to access your profile, or digital twin, as experts call it. Simply check the appropriate boxes on the Gosuslugi portal or elsewhere, and the process is complete.

This is a worrying situation: everyone will know about me.

The government and large companies possess significant amounts of information about citizens. This is not a terrible thing, but such a system requires reliable protection against potential cyberthreats. Ensuring data security should be a priority to prevent attackers from accessing personal information.

What are you trying to say?

Imagine a hacker attack targeting a specific person. If an attacker decides to harm the health of a VIP through food, they can access the system and use artificial intelligence to generate a dish capable of causing anaphylactic shock. This threat highlights the importance of protecting food-related data and systems, as well as the need to develop security measures that can prevent such attacks. Cybersecurity is becoming critical to protecting human health and lives.

This sounds like a plot from science fiction.

It's truly astonishing, but such a scenario must be taken into account if the implementation of this system becomes a reality.

Humans will only eat

Roman Dushkin believes that in the future, people may completely eliminate the cooking process. Artificial intelligence will create customized recipes, and specialized robots will prepare and serve dishes. Humans will be left with only one task: evaluating the culinary creations of machines and providing feedback on their preferences. This technological advancement could significantly change the way we think about cooking and simplify the process of eating, allowing us to focus on the enjoyment of food rather than the preparation itself.

To appreciate how close we are to the future of automation in cooking, check out this video of a robotic kitchen from British company Moley Robotics making soup. It's an impressive example of modern technology that can significantly simplify the cooking process and make it more accessible to everyone.

Moley robots are designed to accurately replicate the culinary techniques of top chefs. Their database contains 5,000 different recipes, but unlike some Russian analogues, they cannot create new dishes.

The robot kitchen is now available for purchase at a price of 250,000 British pounds. Despite the significant cost, the developers have already received hundreds of orders for this innovative device. The product's success confirms the growing interest in the automation of kitchen processes and modern culinary technologies.