Avoiding Bias in AI by Intentional Data Strategy

Have you ever searched online for a photo of a doctor only to notice that what you get back are photos of white males? In order for your AI models to understand various accents, these accents need to be represented in your training data. But even then the job is not done: when the application is live, the data from your actual customers can be used to improve your models and adjust to your customer population. This presentation will outline different steps during your AI application development that ensure that your models are free from bias.

Key Takeaways:

• Data strategy is key to the success of your application
• Avoid bias by making sure all your users are represented in your model
• Live deployment does not mean your job is finished

This session is for... Developers, data scientists, application designers, testers