Improving Digital Transformation in Iraq Using Artificial Intelligence

Two days ago, I visited the Unified National ID Card Issuance Center to issue cards for all family members. There is a specific mechanism for verifying and entering data, part of which is manual and the other supported by a software application. This software was implemented to mitigate the slowness caused by paper-based processes, marking a promising step in the digital transformation of a country like Iraq. However, this is not the happy ending to the story of digital transformation in Iraq.

The queue to reach the data entry clerk is very long and slow, despite the data entry process being supported by a software application designed to speed up the work and increase the productivity of the data entry clerk. Those who were lucky enough to get seats remained seated for a long time without their turn coming up. My children sat on the ground out of boredom and exhaustion. Our privilege of being residents of another province did not help us, as the queue was very slow, and the poor data entry clerk was working with a software application that does not provide the productivity and efficiency supported by current technologies. All this increased my curiosity to understand the mechanism of the software application for data entry, and I began to observe from behind the data entry window how the process worked. I was able to understand the steps of the data entry mechanism, which are as follows:

1. Convert all official documents for each family member into an electronic version using a scanner, including the National ID application, Civil Status ID, Iraqi Nationality Certificate, Residence Card, and supporting documents.
2. Manually enter the information from each individual's Civil Status ID into the software application.
3. Manually compare each individual's information with the information available in the personal records (by comparing with a captured image of the personal records displayed on the application screen).
4. Collect biometric information for each individual.

All the above steps require manual effort from the data entry clerk, making the time to enter the data for a family of five (father, mother, and three children) exceed half an hour. Of course, this is still faster than a non-digitized process, but can the process be made faster using artificial intelligence? Absolutely, especially the second and third steps. The first step can be simplified without AI by allowing the electronic submission of the National ID application, including uploading the required documents electronically during the application process by the citizen or the offices currently handling application creation.

Returning to the second and third steps, these are simply AI functions known as OCR (Optical Character Recognition), which dates back to the 1950s, and computer vision, which involves extracting information from images. The first AI model for reading handwritten numbers was created in 1989 by the French scientist Yann LeCun, who currently serves as Vice President and Chief AI Scientist at Meta (formerly Facebook). The technology, known as Convolutional Neural Networks (CNN), has since evolved significantly, leading to another breakthrough technology known as Transformers, developed by Google scientists through a research paper titled "Attention is All You Need" in 2017. Transformers are the foundation for generative AI like ChatGPT and have also been used in computer vision functions.

Back to our slow steps, the second step can be automated by extracting information using AI functions from the electronic versions of the documents, with the data entry clerk's role limited to verification. The third step can also be shortened by using the same image information extraction technology, converting images of personal records into a database, and automatically verifying the match between the information in the uploaded documents and the information in the personal records database, alerting the data entry clerk only in case of discrepancies.

The expected benefits of using AI-supported systems include speeding up the data entry process and reducing the time required to manually enter the data for a single family from half an hour to a few minutes, minimizing human errors by relying on an automated system for data analysis, and improving the citizen experience by reducing waiting time and making the process smoother and more efficient.

It may be too late to develop the software application for the Unified National ID Card, but digital transformation in Iraq is just beginning. Achieving sustainable digital transformation must be supported by AI, as the world now views applications not supported by AI as akin to working on a mechanical typewriter in the age of electronic computers. There should be AI experts within the IT teams in institutions attempting to implement digital transformation to review current mechanisms and systems, identify where these mechanisms and systems can be supported by AI, and ensure that future systems are AI-supported.