The Silicon Ceiling: How AI Risks Rewiring Workplace Inequality

For decades, companies have worked to dismantle the “glass ceiling” of gender exclusion in the workplace by closing pay gaps, strengthening leadership pipelines and advancing women into decision-making roles. Yet as artificial intelligence reshapes the world of work, a new barrier is quietly taking shape not through intent but through design.

Recent analysis from the International Labour Organization points to a growing structural risk: generative AI is set to disproportionately affect roles predominantly held by women. This is not simply a question of workforce transformation; it poses a direct challenge to Principle 6: the commitment to eliminate discrimination in respect of employment and occupation.

AI systems are exceptionally good at automating the kinds of cognitive, administrative and coordination tasks that have, over time, become concentrated in female-dominated roles. Across sectors, from administrative support to customer-facing functions, these roles form the connective tissue of organizations. They are also among the most exposed to rapid automation.

Without deliberate intervention, this convergence of technological capability and labour market structure risks accelerating inequality rather than reducing it. Digital transformation, left unchecked, could entrench disparities at scale and at speed.

This is where Principle 6 becomes newly relevant. The challenge for companies is no longer limited to fair hiring or representation. It extends to ensuring that technological change does not produce systematically unequal outcomes. If AI adoption leads to disproportionate displacement of women, without pathways for transition, reskilling or progression, then inequality is not being addressed, it is being redesigned.

There is also a clear business case for getting this right. The roles most exposed to AI are often those that hold deep institutional knowledge, organizational memory and relational intelligence. They enable companies to function effectively day-to-day. Treating them as expendable risks eroding not only workforce diversity but operational resilience and long-term performance.

At the same time, expectations are shifting. A new generation of talent is placing increasing weight on how companies balance innovation with responsibility. Organizations seen as advancing technology at the expense of inclusion will find that reputational risk quickly translates into recruitment and retention challenges.

None of this is inevitable. The trajectory of AI in the workplace is not fixed. With the right governance, it can become a force for inclusion rather than exclusion. Creating this future will require a more deliberate approach that embeds gender considerations into how AI is introduced and scaled. This begins with understanding impact before deployment: identifying which roles are most exposed and ensuring that any transformation is matched with meaningful investment in skills and mobility. It requires a shift in mindset, from replacement to augmentation, using AI to remove repetitive tasks while enabling employees to move into higher-value work. It also demands that access to AI literacy and training extends beyond technical teams to the functions where disruption will be most felt.

Crucially, these decisions cannot sit solely with technology teams. AI transformation must be treated as strategic governance issues, with leadership across HR, sustainability and executive management aligned around both business outcomes and social commitments.

With less than four years to deliver on the 2030 Agenda, progress on gender equality must not be taken for granted. The risk is not that AI will replace women; it is that, without intentional action, it will reinforce patterns that have long defined the labour market.

The future of work will be shaped by artificial intelligence. Whether it is more inclusive or more unequal will depend on the choices made now. If the last decade was defined by efforts to break the glass ceiling, this one will be defined by whether we prevent a silicon one from taking its place.