The technology’s unalterable digital trail would reduce fiscal leaks and prevent large-scale corruption draining the public purse. Photo: Supplied
When Albania unveiled “Diella”, an AI-generated minister tasked with overseeing public procurement, last week, it made international headlines. The pitch was simple — an algorithm cannot be bribed, cannot take a meeting in a smoke-filled room, cannot fudge paperwork. By appointing a bot to handle procurement, the Albanian government promised to short-circuit one of the most common avenues of corruption.
The symbolism is striking but it also raises questions that resonate deeply in South Africa. We have lived through the full theatre of state capture and its aftermath. Commissions have exposed wrongdoing in exhaustive detail but accountability has faltered. Technology, whether in the form of artificial intelligence systems or blockchain ledgers, is often presented as a silver bullet. In reality, both can illuminate corruption effectively but neither can punish it. That requires consequence management, something South Africa still struggles to achieve.
Artificial intelligence can process vast amounts of procurement data, welfare payments and tax records and algorithms can detect anomalies and patterns. They can flag suspicious clusters of bids, identify shell companies and highlight unusual payment flows. In theory, this allows scarce investigative resources to be directed toward the most suspicious cases.
The Albanian experiment with an AI “minister” embodies this vision. But AI is not a neutral referee. Algorithms are designed by people and inherit their biases. They can also be manipulated if oversight is weak.
More importantly, an AI system cannot be held accountable. When it decides, or when it is gamed by corrupt inputs, there is no official to cross-examine or prosecute. Responsibility becomes diffuse, spread across coders, officials and overseers. In South Africa, where oversight institutions already struggle, opaque algorithms could produce a new category of unaccountable decision-making.
Blockchain’s appeal lies in its ability to record transactions in a way that is transparent and tamper-resistant. An immutable ledger has obvious uses in contexts like land registration and public procurement, where records have often been manipulated. A pilot project in Khayelitsha used blockchain to capture subsidised property transactions that never made it into the deeds registry, demonstrating how technology can provide secure, verifiable records. In procurement, the same principle could apply — bids, evaluations and awards could be permanently logged and publicly visible, making arbitrary or corrupt changes more difficult to conceal.
International research points to blockchain’s potential role in procurement, asset registries and contract monitoring. The logic is straightforward — if every amendment, every contract and every payment is logged permanently and openly, oversight bodies, journalists and the public can track the process in real time. That kind of transparency reduces the space for fraud.
Yet blockchain is not a cure-all. Its strength is that it preserves records; it does not guarantee their accuracy. A fraudulent contract entered into the system remains fraudulent, only now preserved immutably. The technology makes cover-ups harder, but it cannot stop dishonest officials from feeding false information into the chain. Most importantly, the presence of a reliable record does not compel prosecutors to act. In a political system where consequence management is weak, blockchain might merely produce an incorruptible archive of corruption that goes unpunished.
Comparing AI and blockchain underscores their complementary strengths and their shared limitations. AI can sift through those records to detect suspicious patterns, while blockchain can create records that are harder to erase. Both can reduce opportunities for petty graft and increase the efficiency of oversight.
But neither addresses South Africa’s central weakness — the lack of consequence for those implicated in corruption. The Zondo state capture commission has produced thousands of pages of findings, yet prosecutions have been slow and patchy. Technology cannot replace political will, prosecutorial independence or the enforcement of consequences.
If these technologies are to have value in South Africa, they must be embedded in functioning institutions. AI could be used to triage investigations, provided that decisions remain explainable and subject to human oversight. Blockchain could be introduced in land registries and procurement systems with open public access. Legislation should require transparency in the design and deployment of these tools and interference with them should carry legal penalties.
But without prosecutors who act, auditors who are fearless and politicians who are willing to see cases through, these tools will amount to little more than expensive window dressing.
I write this as someone who has long been a proponent of blockchain technology as a means to curb corruption. Its ability to preserve records and make processes transparent is undeniable. But I have also learned a harder truth — knowing who did it is immaterial in a world with no consequences.
That is the South African dilemma. AI might give us new ways to detect wrongdoing, and blockchain might give us a perfect memory of every transaction, but neither can punish an official who steals. Until our institutions consistently follow through, all the algorithms and ledgers in the world will do is provide clearer, more permanent records of our failures.
Yonela Faba is a PhD candidate in economics at the University of Cape Town, with a research and professional background that spans academia and finance.