Dmitry Masyuk, Director of the Business Group responsible for Yandex’s search and advertisement technologies, highlighted that YandexGPT’s “basic model steadily surpasses” the US ChatGPT 3.5 version in generating responses in Russian.
Furthermore, Masyuk pointed out that the Russian bot often delivers higher-quality answers than ChatGPT 4 in many scenarios. He expressed confidence that it’s only a matter of time before YandexGPT can compete head-to-head with its US counterpart. When compared to LLama 2 7b, another US chatbot developed by Meta, the parent company of Facebook, Masyuk claimed that the Russian bot consistently outperforms it, even when generating responses in English.
Masyuk emphasized that Yandex continuously evaluates its AI product’s progress in comparison to ChatGPT. He acknowledged the challenge of comparing the two systems since they excel in different areas. He posed a rhetorical question, “If a neural network excels in physics tasks, composes fairy tales, and crafts letters to CEOs, but performs slightly worse in other areas, is it ultimately better or worse?”
Yandex introduced its AI system in May, with the purpose of creating chatbots and AI assistants capable of generating, reorganizing, summarizing text-based information, and addressing various business-related tasks. In September, the IT giant unveiled an upgraded version, YandexGPT 2, claiming it can handle a wider range of tasks and offer more precise responses to inquiries.
ChatGPT, developed by OpenAI and launched in 2022, rapidly became the fastest-growing app in history, amassing over 100 million users within two months. This product, a large language model-based chatbot, emulates human conversation and possesses the ability to write and debug computer programs, compose content from fairy tales to student essays, translate text, and answer test questions.
The success of ChatGPT prompted other US tech giants, including Meta, Google, and Microsoft, to initiate their AI product development efforts. The swift advancement of AI technologies has raised concerns about these systems potentially spiraling out of control. In May, the US Congress grilled OpenAI CEO Sam Altman regarding the potential risks associated with the technology.
Altman acknowledged that the technology could cause “significant harm” and had the potential to “go quite wrong.” He particularly expressed concerns about AI’s persuasive capabilities, deeming it one of his “greatest areas of concern.”