While AI can multi-task at speed and do the 'heavy lifting' in the workplace, it cannot empathise, sympathise, imagine or wonder. People should leverage these human characteristics to co-exist with the technology, says Sandra Lehmann, chief actuary at Discovery Vitality.
Speaking to ITWeb on the sidelines of the ITWeb AI Summit 2026, Lehmann acknowledged the value of tools like Claude AI, but added that the next step in AI's evolution will be directed by people.
“AI agents are no longer just advising, they can act," she said. "Things are happening very quickly, but the next step is still very much human – where we want to point it and where we want to apply it.”
Lehmann underlined the significance of raw human qualities like curiosity and imagination, reflecting on her young children. “When I reflect on seeing them, that to me is the raw form of being human. AI doesn’t imagine, AI doesn’t wonder.”
AI like a new employee
Lehmann continued: “AI adoption is like bringing a new employee onboard. A recruit must be guided, mentored and empowered. So too must AI be reviewed and taught."
Operating in a data-intensive environment, humans become overwhelmed by the volume of information and urgency of decisions, she added. AI can act as a silent ‘board member’ if leaders listen to its 'voice’.
“Why not have another voice in the room? It maybe synthetic, but that voice has access to all the information in the world,” she said. “It doesn’t mean you have to believe it, but it’s a voice that isn’t emotional, isn’t part of your business politics and can give you a different lens.”
However, directing AI is not easy, Lehmann suggested, because of the desire to extract value. “It shouldn’t be AI for the sake of AI.”
The human bottleneck
Lehmann said where bottlenecks arise, the reason is inevitably people. “Technology cannot be the excuse. If you do a root cause analysis, it becomes a human issue.”
She believes it is crucial for organisations to always consider human qualities in AI strategies. “It might get to the point where it’s easier to abdicate decision-making to the machine, but that’s a dangerous path to go down.”
Share
