At the CIO of the Future Summit in Dar es Salaam, tech leaders discussed AI's impact on daily operations. Among the panelists was Imani Mwanri, head of IT at Reliance Insurance, who urged tech leaders to embrace AI while warning against abandoning human oversight.
Imani Mwanri, head of IT at Reliance Insurance, has witnessed firsthand the benefits a company can draw from digital tools. During the CIO of the Future Summit in Dar es Salaam, he demonstrated how his company uses AI to transform customer touchpoints while maintaining essential human checks .
“Say you need a quotation. In places where we can do end-to-end, we have authentication processes to ensure that integrity of data is maintained. And we do that using AI,” he said.
However, he also emphasised that sometimes AI can be limited, necessitating human involvement.
“Sometimes we need a human factor to decide, so we stop. Sure, we can interact with the AI up to the end, but we'll drop an email to the user networks to say ‘you need to finalise the process. You need to decide,” he said.
The intelligence balance
At Reliance, the use of AI is guided by Imani’s belief that for AI to work well, there must be a balance between technical execution and human input. He explained that true enterprise automation requires distinguishing between human intelligence and artificial intelligence.
“AI includes the machine and the robotic process automation but it is made by humans. So you cannot leave aside human intelligence. The system can decide the data because of the experience it has developed but the human knows the processes better. I believe we must be able to separate between the two because it is critical.”
To further emphasise the importance of the human factor, he explained that critical evaluation and system monitoring remain uniquely human capabilities.
"What you cannot measure, you cannot monitor, and measuring intelligence requires critical thinking that only a human mind can supply.”
This, he argued, is where AI's limitations become most apparent. AI systems operate on the commands and prompts they are given, and if a user enters an incorrect prompt, the system has no way of recognising or correcting the error on its own.
"AI cannot come and apologise. If a mistake is made and needs to be acknowledged or corrected, particularly with a customer, that responsibility falls to a person, not the algorithm.”
Even though he advocates for AI adoption across key operations,he remains cautious regarding cybersecurity risks and system unpredictability. As a result, at Reliance Insurance, the company is cautious on how AI is used and the data that is fed to it.
“We invest in creating user awareness but we also ask ourselves what would happen if one day AI works the way we wouldn’t want. For this reason, we keep evolving and moving ahead and we also create mechanisms to ensure that our operations would still continue in such a situation.”
