Specialists from Saudi Arabia and China have developed an artificial intelligence (AI) system focused on the Arabic language.
The collaboration involved scientists from King Abdullah University of Science and Technology, the Chinese University of Hong Kong and the Shenzhen Big Data Research Institute.
A large language model called AceGPT is created as an open source solution based on Llama 2 from Meta Corporation.
“The tool not only takes into account the unique syntactic complexities of the Arabic language, but also ensures cultural sensitivity and alignment with local values,” the description says.
The model is designed and trained to work as an AI assistant for Arabic-speaking users. However, the researchers warned that in other languages it may not generate “satisfactory” responses to queries.
The developers also noted that AceGPT potentially has shortcomings. They included possible abuses such as handling confidential information, creating harmful content and supporting disinformation.
The researchers stressed that they did not conduct exhaustive security checks and advised users to “exercise caution.”
“We cannot overemphasize the need for responsible and intelligent use of our model. Moreover, our assessments were mainly based on data from open sources and on data obtained during the work,” said the international team.
Let’s remember that at the end of September Meta presented a universal set of AI tools, created including on the basis of Llama 2.