Short answer

Actively audit AI-generated imagery for demographic representation and bias, and consider manual adjustments or alternative image sources to ensure inclusivity and accuracy.

Field: Innovation & Design
Source: Laryngoscope Investigative Otolaryngology (2025)
Method: Comparative analysis and statistical comparison
Sample: 1740 portraits (580 per platform)
Evidence: Strong effect

AI text-to-image platforms exhibit significant biases, underrepresenting women and racial minorities while overemphasizing White males in professional contexts, mirroring and potentially amplifying existing societal stereotypes. This innovation & design research insight is drawn from a 2025 study published in Laryngoscope Investigative Otolaryngology. Using Comparative analysis and statistical comparison with 1740 portraits (580 per platform), researchers explored how this design variable affects real-world outcomes. The key design takeaway: Actively audit AI-generated imagery for demographic representation and bias, and consider manual adjustments or alternative image sources to ensure inclusivity and accuracy.

Study

Innovation & DesignNew This WeekStrong effect

AI Image Generators Perpetuate Gender and Racial Stereotypes in Professional Portrayals

Laryngoscope Investigative Otolaryngology · 2025

Key Findings

01AI platforms demonstrated significant racial and gender biases in generated portraits.
02Female and racial minority representation was consistently lower than actual workforce demographics.
03White males were disproportionately overrepresented across the tested platforms.

Application

Design takeaway

Actively audit AI-generated imagery for demographic representation and bias, and consider manual adjustments or alternative image sources to ensure inclusivity and accuracy.

How to apply

When using AI image generation for marketing materials, website visuals, or any design collateral, explicitly prompt for diverse representation and then critically review the output against real-world demographics.

Project actions

01When using AI for image generation in your design project, be aware that the results might not be diverse.
02Always check the AI's output to see if it accurately represents different groups of people.

Method & Evidence

AimTo investigate the extent to which AI text-to-image platforms reflect or amplify gender and racial biases when generating professional portraits, comparing outputs to real-world workforce demographics.

MethodComparative analysis and statistical comparison

ProcedureThree AI text-to-image platforms were prompted to generate 580 portrait photos each for otolaryngologists, using various descriptive categories. Two reviewers then characterized the gender and race of the generated portraits. The demographic distribution of the AI outputs was statistically compared against known workforce demographics.

Sample1740 portraits (580 per platform)

ContextProfessional imagery generation by AI, specifically in the medical field (otolaryngology).

Variables

IVAI text-to-image platforms (DALL-E3, Runway, Midjourney), descriptive prompts (personality traits, fellowship, academic rank).

DVDemographic representation (gender, race) of generated portraits.

CVThe profession being depicted (otolaryngologists), the number of images generated per platform.

Strengths & Limitations

Strengths

+Utilized multiple AI platforms for a broader comparison.
+Employed human reviewers for demographic characterization, adding a layer of interpretation.

Limitations

The specific biases found might be unique to the AI models and prompts used in this study and may not apply universally to all AI image generation.

Reliability & validity

Reliability could be improved by using multiple reviewers and ensuring consistent criteria for demographic classification. Validity is supported by comparing AI outputs to established demographic data, but the interpretation of 'race' and 'gender' by AI and reviewers can introduce subjectivity.

Think critically

How can designers actively counteract or mitigate the biases embedded within AI image generation tools to ensure their design outputs are inclusive and equitable?

Design Principles

"Critically evaluate AI-generated content for bias and ensure it aligns with desired representational goals."

As AI tools become more integrated into design workflows, understanding their inherent biases is crucial for creating equitable and representative outputs. Designers must be aware that AI-generated imagery may not accurately reflect diverse populations, leading to the perpetuation of harmful stereotypes if not critically reviewed.

What This Means for Your Design

AI tools that make pictures from words can be biased, often showing more white men than women or people of color, which isn't what the real world looks like.

How to use in your project

1.Reference this study when discussing the limitations of AI tools in your design process or when justifying the need for manual image curation.

Add to My Project

Quick Cite

Paragraph starter

The integration of AI text-to-image platforms into design practice necessitates a critical awareness of their inherent biases. Research indicates that these tools often perpetuate gender and racial stereotypes, underrepresenting minority groups and overemphasizing dominant demographics, which can lead to the creation of non-representative and potentially harmful visual content.

Source

Laryngoscope Investigative Otolaryngology

Representation of Demographics in Otolaryngology by Artificial Intelligence Text‐to‐Image Platforms

journal · 2025

View source

Questions About This Research

What does the research say about ai image generators perpetuate gender and racial stereotypes in professional portrayals?: Actively audit AI-generated imagery for demographic representation and bias, and consider manual adjustments or alternative image sources to ensure inclusivity and accuracy. Evidence: Laryngoscope Investigative Otolaryngology (2025).
Why does "AI Image Generators Perpetuate Gender and Racial Stereotypes in Professional Portrayals" matter for design?: As AI tools become more integrated into design workflows, understanding their inherent biases is crucial for creating equitable and representative outputs. Designers must be aware that AI-generated imagery may not accurately reflect diverse populations, leading to the perpetuation of harmful stereotypes if not critically reviewed.
How can designers apply this research?: Actively audit AI-generated imagery for demographic representation and bias, and consider manual adjustments or alternative image sources to ensure inclusivity and accuracy.
What were the main findings?: AI platforms demonstrated significant racial and gender biases in generated portraits.. Female and racial minority representation was consistently lower than actual workforce demographics.. White males were disproportionately overrepresented across the tested platforms.
What research method was used?: Comparative analysis and statistical comparison with 1740 portraits (580 per platform).
How strong is the evidence?: Evidence strength is rated Strong effect, based on a 2025 journal from Laryngoscope Investigative Otolaryngology.
What should I do differently in my next project?: When using AI image generation for marketing materials, website visuals, or any design collateral, explicitly prompt for diverse representation and then critically review the output against real-world demographics.
What are the limitations?: Biases may vary across different AI platforms, prompt engineering techniques, and specific professional fields beyond otolaryngology.