Information Technologies Overview For Creating Intelligent Assistants In Culinary
Abstract. The general concept of an intelligent assistant and it’s corresponding basic functionality are considered. Functions inherent in the field of cooking are distinguished: management of food production and preservation processes, recipes generation, optimal recipes selection, menu creation, diet planning, monitoring and adjustment, shopping lists formation. The general structural diagram of a culinary intelligent assistant is presented.
An intelligent assistant (AI) is an interactive program designed to assist users in solving various tasks, such as providing information, automating routine processes, providing personal assistance, voice interaction, integrating with other programs, and creating unique content. It is the latter function that distinguishes the modern intelligent assistant from its predecessors, such as chatbots and digital personal assistants.
Chatbots typically solve the problems of automating/accelerating the execution of typical/standard requests. In the case when the chatbot does not find a standard answer to the request, the user is usually switched to interaction with a real person. Initially, chatbots were non-adaptive systems, as they were only able to answer standard questions. A digital personal assistant is an advanced version of a chatbot that can read the context and insert information from various sources into it. An intelligent assistant creates unique content using generative artificial intelligence methods.
The main drivers of AI implementation in cooking are the following main factors. Firstly, it is the implementation of Internet of Things (IoT) technology both in personal kitchens and in the corresponding professional environment, which provides a completely different level of overall process efficiency. Secondly, it is the use of machine learning and artificial intelligence algorithms to support personalized decision-making. For example, adapting recipes to personal preferences, developing a special menu that takes into account dietary requirements, product availability, cooking time, etc.
Intelligent assistants in this field can be conditionally divided into several main groups depending on the tasks they solve, namely: managing the processes of food production and preservation, generating recipes, selecting optimal recipes, creating menus, planning, monitoring and adjusting diets, creating shopping lists, etc.
The basic steps that an AI implements in its work are receiving a message from the user, processing the message, obtaining information, and generating a response. Let’s consider the specifics of implementing each mentioned step.
During the first step, you need to send a message (request) directly to the assistant. This initial message is usually sent by the user in one of the following basic ways: voice command, text message. Additionally, physical parameters of the external environment can be transmitted, such as temperature and humidity. This message must be processed and submitted to the assistant’s input in a formalized form. And if physical parameters are read from special devices in a formalized form and can be directly transmitted to the system, then text or voice messages require additional processing.
In the second step, AI uses natural language processing methods - this is a direction in machine learning dedicated to the recognition, generation and processing of spoken and written human language. If the request is generated as a voice message, the following steps are implemented: recording human speech with an audio device; converting words from audio to written text. In the information retrieval step, the text is parsed into components, the context and purpose of the request are understood. Finally, a set of actions is determined that must be performed in response to the initial request.
In general, the structural diagram of a culinary AI should have the following main components:
- generative AI module (generates recipes, budgets, plans and recommendations; uses NLP for user interface);
- IoT integration module (receives and analyzes data from smart devices; interacts with home systems for automation);
- personalization module (takes into account the user’s health, preferences and lifestyle; works with wearable devices and health applications);
- financial and social module (helps plan expenses and purchase products taking into account the budget. suggests recipes or events to improve the quality of social life);
- ethical module (monitors data confidentiality and compliance of the system with ethical standards).