Professor Reframes AI Job Debate Around Power

Professor Linda du Plessis, senior deputy vice-chancellor at North-West University, argues the wrong question about automation is whether AI will take jobs. The correct question is who controls AI and whose interests guide deployment. Du Plessis warns a small group that controls leading AI systems, guided by corporate incentives and political choices, will shape labor outcomes far more than technical capability.

The argument is governance-focused rather than model-focused. Key technical realities practitioners should keep in mind on the governance axis are:

• •Concentration of development resources and production-grade models within a few organizations creates single points of incentive-driven decision making.
• •Design choices, deployment policies, and economic incentives determine whether systems augment work or replace roles.
• •Transparency, auditability, and participatory design are practical levers that change outcomes for workers and communities.

This perspective places AI risk assessment in the realm of institutional economics and public policy rather than pure engineering. It aligns with recent debates about model access, platform power, and the need for public-interest compute and datasets. The call for universities to act as moral custodians echoes a broader push for civic institutions to provide independent verification, education, and governance frameworks. For practitioners, that means policy engagement, designing for augmentation, and building features that enable oversight and redress, not only performance gains.

Whether public universities and research institutions obtain stronger roles in procurement, auditing, and civic education on AI. Also watch corporate governance moves: will firms embed worker protections, augmentative design patterns, and transparency practices, or will market incentives drive further concentration and displacement?

The piece reframes workforce impact as a political and governance choice, not an inevitability of algorithms. Practitioners should treat architecture and deployment decisions as policy levers with measurable labor and social consequences.