LANGUAGE SPACE OF NEURAL NETWORKS: FEATURES AND DIFFERENCES FROM NATURAL LANGUAGE
Background. Modern society is developing in a world of information overload and digitalization, which has led to a need for artificial intelligence (AI) to help analyze and process information more efficiently. Most AI models use natural language processing, and while they can imitate human speech,...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Science and Innovation Center Publishing House
2024-12-01
|
| Series: | Sovremennye Issledovaniâ Socialʹnyh Problem |
| Subjects: | |
| Online Access: | https://soc-journal.ru/jour/index.php/mssi/article/view/440 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Background. Modern society is developing in a world of information overload and digitalization, which has led to a need for artificial intelligence (AI) to help analyze and process information more efficiently. Most AI models use natural language processing, and while they can imitate human speech, they generate texts based on templates set by humans. This creates an online language space that is unique to AI, with its own characteristics due to its isolation from human interaction.
Purpose – analysis of lexical and grammatical features of artificial intelligence language and comparison with human language.
Materials and methods. To solve the tasks set in the research, we used various methods of analysis: observation, description, and the comparative method. The article is based on an analysis of texts generated by neural networks such as ChatGPT, Rytr, Smodin, and ChatSonic, which were created at the request of the authors.
Results. The functioning of artificial intelligence (AI) is based on simulating the thought processes that are inherent in the human mind. This imitation is made possible by the development of language modules by AI specialists for use by neural networks in natural language processing (NLP). The language space of AI, limited by the input data from these modules, can generate texts that are practically indistinguishable from natural language, although they have features that are determined by the conditions under which they are created. These features include attachment to keywords in the task, potential loss of the main idea, inconsistency with required style, repetition of similar ideas, lack of specific facts or personal experiences, and limitations in vocabulary, linguistic patterns, and grammatical structures. The current inability of artificial intelligence (AI) to go beyond its current linguistic capabilities is due to a lack of a concept such as consciousness in neural networks. This allows for the accumulation of experience and the ability to think with insights, which are essential for AI to progress beyond its current limitations. |
|---|---|
| ISSN: | 2077-1770 2218-7405 |