AI Tech Suite

See You There

Click to visit website

About

See You There is an innovative generative AI project launched by Black & Abroad for the 2024 travel season. The tool generates personalized photo albums featuring past guests at upcoming destinations based on their previous images. This project uncovered significant racial bias in AI image generation, highlighting that the AI models often created images of Black guests with lighter skin, non-Black features, and misleading representations. Through hundreds of generative tests, the initiative strives to shed light on algorithms' biases and advocate for more inclusive and accurate representation of Black travelers in the AI-driven travel photography domain. The project emphasizes using advanced AI technology to enhance travel experiences while addressing the systemic issues of bias prevalent in AI modeling.

Features

• generates personalized photo albums for travelers

• highlights bias in ai-generated images

• prompts exploration of racial representation in ai

• interactive bias reporting system

• data collection for improving ai models

FAQs

What is an AI prompt?

An AI prompt is instructions given to an AI model to generate desired output. For this project, prompts directed the models to see guests at various destinations.

Did you get permission to use people’s photos?

Yes, guests filled out waivers for their past images to be used in this project.

What generative AI Models did you use?

The project used a widely available generative AI platform, but it aims to address bias systemic in society, culture, and technology.

How many tests did you run?

The project operated over 4 months from December 2023 to March 2024, running hundreds of generative tests.

What factors led to problematic images being generated?

We identified geographic destination, specific activities, prompt specificity, and training data quality as key factors impacting image accuracy.

How did you change your prompts to achieve accurate photos?

Adding detail to prompts improved results significantly. Specificity in prompts led to better accuracy.

If generative AI is so racially biased, why did you use it?

Generative AI offers potential for innovative experiences, and this project aims to address bias to create an inclusive vision of travel.

Why doesn't the generated photo look exactly like the real guest?

Generated photos vary due to the inherent bias in AI models, altering features, body types, and more. They are not meant to be pixel-perfect representations.