GREGORY LEE: Change of guard at home affairs a chance to deploy wise digitalisation

Just less than a year ago I was kicking myself for missing the chance to respond publicly to something home affairs minister Aaron Motsoaledi had said. He had just delivered one of the daftest public whinges ever delivered by a government minister — and that is a competitive category.

The nature of his bleat du jour was that home affairs would never clear the epic backlog of visa applications without bloating its workforce by a staggering margin.

I recently noted that the aforementioned and ongoing visa backlog, especially in the scarce skills category, has become a major economic inhibitor for SA. Admittedly, we have seen some movement in the work visas arena. However, in too many respects home affairs remains a dumpster fire under our economy’s braaivleis.

Personally, our family nearly had to emigrate because of a severely delayed permanent residency application for my foreign wife that had become unbearable personally and economically.

Now that the DA’s Leon Schriber is taking over this portfolio, maybe the monomaniac obsession with immigrants can give way somewhat in favour of a move to genuine efficiencies. And, no, we don’t need to swell the bodies behind the desks. By all means, add people if truly required. But we can do a lot with digitalisation of government services in general, and home affairs in particular.

This is why Motsoaledi’s statement was so misplaced — there remains a wealth of possibilities offered by robotic process automation and a little sprinkling of (sensible) artificial intelligence (AI) that would be more fiscally feasible than an enormous hiring splurge.

However, well-padded, gravy-stained government fingers have been badly scorched before by ill-advised huge digitalisation attempts, ranging from the e-Natis pre-eclampsia to home affairs’ own aborted attempt to implement an automated biometric identification system.

So, what can we achieve in good time and without huge cost to digitally improve home affairs?

Plan and test big integrated platforms slowly for scalability

The graveyard of government attempts to implement huge platforms from a standing start far exceeds the podium of successes. This is not to say there are no successes. Within home affairs I would count the passport and ID card booking system as a highly qualified success, in that it achieves most of its goals. That is, allowing you to digitally prefill a form and book and appointment with only intermittent outages of service.

However, even this citizen-orientated system is highly limited and often stubbornly hard to use. Children’s passports must be applied for through truncated means. You can only book with banks there — there is a different, parallel system for booking with home affairs branches. So, lots to do still.

There are multiple issues with trying to build integrated customer-focused systems such as these. Scalability is critical and often not properly accounted for. A system can work fine in the testing phase with only a few proof-of-concept users, but how does it fare when thousands of users are trying to log in simultaneously?

Have you considered how the system will perform once it needs to search through hundreds of thousands of records, and have you catered for the storage of all those records? These are the questions that have not been well-answered in the past.

Also, the vast digital divide in SA does not necessarily make customer-focused platforms and apps the priority for now. While the future should undoubtably involve such endeavours, there are lower-hanging, cheaper and quicker fruit that the revamped department could consider.

One request: please don’t make us fill in the same piece of data multiple times in a process. Can we only give our name once, our email once, and so on? And, even better, can we only do it once ever and have the system prefill the fields with our central records? Such bliss just thinking about it!

Deploy robotic process automation

Software bots that can undertake myriad software-based tasks automatically or with human intervention are perfect for improving the kinds of back-office processes with which home affairs is embroiled. The department is undoubtedly using this technology already, but anyone who has recently interacted with them knows there is room for improvement.

Robotic process automation can cheaply and effortlessly help bridge the gaps between stand-alone systems, move processes along and prompt human decisionmakers, fetch and fill in data quickly, populate and edit data sets, and far more. For instance, a birth certificate should be open to immediate generation using a bot that can populate the request, check and fetch supplementary records, create the document and print it.

Indeed, there has been a project under way, but announced three years ago at the staggering cost of R2.4bn. Anyone in the automation field will tell you that is dramatically overpriced in modern terms. There remains a lot of runway for the robotic process automation plane.

Use AuditBots

Perhaps one of the most useful options for robotic process automation is one that citizens would not even see: AuditBots. These pieces of code are designed to crawl through databases on an ongoing basis — as often as several times a day, but usually less — to search for anomalies and report it.

Someone has been registered as having been married? Can a daily AuditBot pick this up and quietly check in the background for possible signs that this is a fraudulent marriage, as has happened too often before? Automatically inform the citizen that they are suddenly hitched?

Wise AI

Don’t try to throw AI at everything, especially advanced AI such as generative solutions. Well-established AI algorithms exist that are highly functional in fields that are useful to home affairs, such as automatic ID recognition, language detection, translation, and so on. In my extensive tests over the past year of the OpenAI language tech underlying Microsoft’s Power Automate functions I have found it increasingly good at detecting a variety of SA languages (but not all) and translating them. This sort of thing, alone, is where we can obviously add huge functionality quickly.

By the way, let’s not for a second believe Schriber can claim any prominence over the ANC’s Motsoaledi in improving the department. Schreiber will need to prove himself and show his digital chops before anyone should be impressed. But as we like to say, a change is as good as a holiday — so long as that holiday is not a polluted KwaZulu-Natal beach.

• Lee, an associate professor of digital business at Wits Business School, is co-author and editor of several books on the digital economy.