AI in Iraq: why it has to get Kurdish and Arabic right

AI is moving quickly into Iraqi organisations. But most of these systems were built for English, and they have real problems with Kurdish and Arabic.

Common problems

1. Wrong letters: Arabic «ي» and «ك» used in place of Kurdish «ی» and «ک», which breaks search and sorting in databases.
2. Text direction: ordering breaks when numbers and English words mix with right-to-left text.
3. Literal translation: output reads as word-for-word translation rather than natural language.
4. Data abroad: sensitive information is sent to servers outside the country.

The fixes

• Models and language rules improved with local data
• Automatic checks for spelling and letters
• On-premise systems such as AI Box for sensitive data

NetGrow works on these solutions from Erbil for organisations across Iraq.