Generative AI has become the default tool for creating professional content — résumés, headshots, promotional imagery, leadership assets, even children’s homework. It is now shaping the visual language of work and leadership at a global scale.
What does this mean for diversity, equity and inclusion?
Look at the images AI at Meta produces when you prompt it for a “LinkedIn profile picture of a…
Entrepreneur
Lawyer
Banker
Doctor
Architect
Authority on leadership
Across these examples, the pattern is unmistakable—overwhelmingly white, male, able-bodied, not so mature in age. Even when cartoons are generated, the stereotypes become exaggerated, not diminished.
These patterns matter. AI isn’t just producing images — it’s shaping expectations. And the risks are far-reaching and deeply structural.
The Real Risk: AI Will Socialise a Generation Into Stereotypes
AI is teaching children, teenagers, and early-career professionals what “success,” “leadership,” and “professionalism” look like.
1. AI becomes a teacher — but a biased one.
When a child asks an AI tool for a picture of “a leader,” and receives a near-uniform set of older white men in suits, the message is clear: leadership looks like this, and not like you.
2. Stereotypes become automated and normalised.
The more users generate images, the more the model reinforces the pattern: Men in leadership roles. Women in supportive roles. People of colour underrepresented or absent.
This hardens old stereotypes into algorithmic defaults — far harder to dismantle than human prejudice.
3. Children internalise the limits AI shows them.
A young girl searching for “a CEO” or “a scientist” may see few images of women. Children from culturally diverse backgrounds may not see themselves represented at all. What they don’t see, they may not believe they can become.
4. Psychological impacts intensify inequalities.
Research has long shown that:
- Representation shapes aspirations
- Stereotypes affect confidence and performance
- Role models impact long-term career pathways
If AI systematically excludes certain groups, the long-term socio-economic impact will be enormous.
5. Workplace DEI progress is quietly reversed.
Organisations have spent decades:
- Increasing representation in leadership
- Challenging biased stereotypes
- Building inclusive cultures
- Disrupting narrow definitions of professionalism
Generative AI risks reversing those gains by re-normalising old biases at scale — often without anyone noticing.
6. AI-driven hiring tools may further discriminate.
When biased image-generation co-exists with biased résumé screening, performance prediction tools, or “professional photo enhancers,” the entire employment lifecycle becomes a biased pipeline.
We risk creating a self-reinforcing feedback loop where the groups AI sees as “leaders” are those it continues to select and promote.
Why This Is Happening: The Structural Sources of AI Bias
1. The datasets are biased.
Most models are trained on existing online imagery and text, which reflect decades of systemic inequalities:
- Overrepresentation of white men in leadership photos
- Underrepresentation of women in STEM
- Stereotypical imagery of race, culture, and professions
AI doesn’t create bias — it multiplies it.
2. The teams building AI are not diverse enough.
When designers, engineers, and product leaders come from a narrow demographic group, blind spots go unchallenged and biases remain invisible.
3. Algorithms optimise for familiarity, not fairness.
Engagement-driven systems often prioritise traditional, “safe,” stereotypical outputs. What feels familiar gets surfaced more.
4. There is no global regulatory framework.
We have safety standards for food, medicine, vehicles, and advertising — but not for a technology rewriting social norms.
5. AI companies are moving faster than governments.
The scale, speed, and accessibility of generative AI outpace policy-making.
The Consequences: What We Stand To Lose
Without intervention, we risk:
- Erosion of DEI gains across every sector
- Entrenched racial and gender stereotypes presented as “neutral” AI outputs
- Widening of the confidence gap for underrepresented groups
- Bias in hiring, performance assessments, and professional branding
- Cultural homogenisation, where Western norms dominate global content
- Long-term harm to children who form self-beliefs around AI-generated images
- Reduced economic inclusion, as biases shape opportunity pathways
This is not speculative. It is already happening — visibly, repeatedly, and quietly.
The Path Forward: Concrete Solutions
Below are the critical, practical steps that Meta, AI companies, governments, and organisations must take to safeguard inclusion.
Solutions for Meta, OpenAI, Google, and Other AI Providers
1. Build diverse, representative training datasets
Models must include:
- Balanced representation across race, gender, body type, age, culture, and ability
- Over-sampling of historically underrepresented groups
- Removal of biased or stereotypical historical imagery
2. Create “bias interruption” layers in generation tools
This means:
- Diverse outputs by default for roles like CEO, doctor, athlete, founder, leader
- Alerts when a prompt is likely to generate biased outputs
- Prompts offering more inclusive alternatives (“Would you like a diverse set of representations?”)
3. Conduct mandatory bias tests on professional role prompts
Every major role should be tested for:
- Gender distribution
- Racial distribution
- Stereotype reinforcement
- Cultural bias
Results must be made transparent.
4. Employ diverse AI development teams
Representation among creators reduces blind spots in the system.
5. Publish annual AI bias audits
Like financial audits but focused on:
- Disparities in outputs
- Examples of stereotype reproduction
- Steps for corrective action
These should be public and independently verified.
Solutions for Governments and Regulators
1. Establish global standards for AI fairness
Similar to cybersecurity or data privacy standards.
2. Require transparency in training datasets
Not the full data — but disclosures about balance, representational gaps, and sources.
3. Regulate AI used in employment, education, and media
Bias testing should be mandatory before tools are deployed.
4. Create an independent AI Bias Safety Commission
A regulator that:
- Tests algorithms
- Monitors outputs
- Investigates complaints
- Issues compliance certifications
5. Protect children specifically
Introduce safeguards for:
- School-based AI tools
- Image-generation systems
- AI content designed for minors
Children should not be exposed to biased systems without protections.
Solutions for Organisations
1. Create an AI Governance Framework with DEI at its core
This should include:
- Mandatory bias checks before publishing AI-generated content
- Requirements to use approved tools with bias-mitigation features
- Clear rules against using AI images for hiring or selection
2. Train employees on AI bias literacy
Staff need to understand:
- How bias appears
- How to check for it
- How to correct outputs
- When not to use AI
3. Conduct DEI assessments on all AI-assisted branding
Before posting:
- Leadership imagery
- Recruitment campaigns
- Website visuals
- Corporate storytelling
A DEI gate-check must be applied.
4. Appoint internal AI ethics leads or committees
These teams monitor:
- Vendor selection
- Data use
- Risks in marketing, HR, and leadership communications
- Ethical compliance
The Stakes Could Not Be Higher
AI will influence how billions of people across all ages see themselves and others. It will shape narratives of leadership, competence, value, and belonging.
If we fail to intervene, we risk embedding old hierarchies into the next generation’s consciousness — invisibly, automatically, and globally.
But if we do act — with urgency, courage, and inclusion — AI can become one of the greatest tools for expanding representation, challenging stereotypes, and accelerating equity.
The question is not whether AI will reshape society. The question is who it will include — and who it will erase — if we don’t get this right.
Related Reading:
Why Has Women’s Visibility Collapsed on LinkedIn?
Stereobots: When Chatbots Get Typecast — and How We Can Recode Gender in AI
The Intersection of AI and DEI
AI and Organisational Culture: Lessons from Deloitte on Trust, Transparency, and the Human Factor
Guarding Against AI Bias in Talent Management: Building Fairer, Smarter Workplaces
Integrating AI into Your Next DEI Strategy