Building Fairer and More Explainable Solutions with AI Verify

Dell Technologies is a leading American multinational technology company that develops, manufactures, sells, and supports computers and related products and services. The company helps to shape the future of digital transformation by enabling businesses and consumers worldwide to innovate, grow, and succeed.

Dell tested AI Verify on:

• AI Model: Light GBM AI model
• Use Case: This AI model was built for an institution conducting frequent donation drives. The model helps predict the likelihood of recipients contributing funds, considering factors such as their donation history and demographic attributes. Recognising the variability in donor behaviour, where not all are repeat donors and new donors have different propensities for giving, the focus was on identifying prospects – both new and existing – to receive targeted mail. This approach ensures a higher probability of campaign success, concurrently minimising costs associated with mail distribution for the organisation.

Dell Technologies’ positive experience with AI Verify

Utilising sensitive demographic data like race, gender, and age as input features poses the risk of machine learning models associating these factors with specific outcomes. This could potentially result in unfair bias being directed towards marginalised groups. As an organisation that strives to provide trustworthy AI solutions, AI Verify serves as a good starting point in evaluating fairness and performance across different use cases. The relevant metrics churned by AI Verify has helped the Dell team to identify potential biases for the existing and future models.

Dell believes that demonstrating trustworthy AI is important because it helps to build and maintain customer confidence, to reduce compliance costs, and most importantly to mitigate the risk of harm to end users. The AI Verify toolkit provides a unified framework for organisations to implement honest, fair and equitable systems.