Credit：TMTPost

iFlytek announced its AI-powered large language model (LLM) Spark Desk was accessible to the public, and users can download the model on any major App store for registration and usage. LLM is the AI technology that empowers ChatGPT, the most popular AI chatbot developed by OpenAi since late 2022. The release made iFlytek one of the first listed Chinese companies to launch their AI offerings to the public.

iFlytek unveiled Spark Desk on May 6, and the chairman Liu Qingfeng said the first batch of over 3,000 enterprises from 36 industries will deploy the model as the launch. The multitask model covers key capabilities including AI-generated content, multi-language understanding, logical reasoning, mathematics and coding, according to Liu. "We aim to exceed ChatGPT's capabilities in Chinese language by October 24 and will achieve a similar level to it in English," Liu said.

About three months later, the firm rolled out on August 15 the upgraded Spark Desk 2.0. The new version has improved coding capabilities in terms of code generation, completion, error correction, explanation and unit testing generation. With upgraded multi-modal capabilities, it can read and analyze pictures, generate photos, and create pictures and short videos based on ancient poems.

iFlytek’s Spark Desk was ranked first followed by models from companies including Baidu, SenseTime and Alibaba, according to the domestic mainstream models evaluation list released by Xinhua News Agency's Chinese Enterprise Development Research Center in August. Specifically, Spark Desk took the first in the two dimensions of intelligence index and tool efficiency index among the four evaluation dimensions designed by the research center.

Spark Desk’s release to public came after the public launch of LLMs first got the nod from Chinese government. A group of LLMs developed by a total of 11 Chinese companies or institutions including Baidu, SenseTime, ByteDance obtained approval in late August. Eight LLMs such as Baidu’s  Wenxinyiyan, Douyin’s Yunque were available for the public on August 31. Baidu also had a large-scale launch of its chatbot Ernie to the public that day, five months after the bot’s inception.

The Cyberspace Administration of China (CAC) introduced in July the Interim Measures to Regulate Generative AI Services (“Measures”), the first of rules on AI-based generative contents and services in China since the launch of ChatGPT. The Measures, effective on August, 15, require companies to submit security assessments and receive clearance before they offer generative AI offerings to the public. Companies also have to apply for registration of their algorithms with relevant authorities prior to release of mass-market AI products and report any changes of the algorithms.

Providers of generative AI contents shall take responsibility as the creator of the content produced by the product, and they have to protect personal information if their possessions involve individual users’ information, according to the Measures. The Measures stipulate AI-powered content shall adhere to socialist values.  The content produced by generative AI should align with core socialist values. No content that attempts to overthrow the government, the socialist system, or foment to split the country, hurt the national unity, preach terrorism, extremism, ethnic hatred, ethnic discrimination, violence, obscene and pornographic information or content that may disrupt the economic and social order should be allowed, reads Article 4.1 of the Measures.

763人已贊賞 >

敬原創(chuàng)，有鈦度，得贊賞

贊賞支持

0

0

0

掃描下載App