SolasAI, a provider of market-proven algorithmic fairness AI software, this week announced enhancements to the SolasAI Bias Explainability & Mitigation Library to produce fairness results faster and more efficiently while enabling companies to safely and quickly innovate with artificial intelligence. The update reduces fairness results processing time from 48 hours to within one hour.
SolasAI has improved feature selection results of the Bias Explainability & Mitigation Library and their competitors with fewer computational resources necessary. The company also solved coding issues with larger data sets and made significant speed improvements, creating instantaneous advances in team collaboration and effectiveness while slashing turnaround time for producing results.
“What we’ve created in this latest update can only be replicated by top-five banks, and smaller banks and fintechs do not have the budget to do what we’ve been able to achieve,” said Nicholas Schmidt, co-founder and chief technology and innovation officer at SolasAI. “This creates a situation where smaller companies can be just as, if not more, effective than larger competition based on what’s available in the market. This is part of our vision to democratize AI and reduce bias and discrimination at all levels of business, and we’re one step closer to achieving this.”
SolasAI conducted private pilot testing with focus groups. Participating companies ranged from Fortune 100 healthcare and financial service companies constrained by computing power to AI-driven lending fintechs requiring optimized analysis to curb cloud computing costs and support legacy environments. Following the update, the groups saw reduced memory requirements for feature selections, leading to faster and more robust results without needing to improve hardware.
Smaller businesses commonly have two more considerable shortcomings that hold back performance compared to larger competitors: resources and access. Companies often don’t have the required features in one package or host an automated or curation system that can organize data. Without resources to hire employees with high levels of data and fairness expertise, they keep models simple and easy to explain. Some companies may outsource or use third-party AI-driven services for models, so they rely on others to handle their AI modelling.
The Bias Explainability & Mitigation Library, SolasAI’s paid testing and mitigation tool, gives smaller companies a unified, curated and easy-to-use library to handle more complex AI models without a robust investment in programming, data science and algorithmic fairness. The library also allows smaller banks, credit unions and healthcare providers to test and resolve issues with third-party models at an affordable price.
“SolasAI is about innovating to help reduce discrimination for as many people as possible, and this latest update places us in the driver’s seat to lead that effort,” said Larry Bradley, CEO at SolasAI. “Smaller businesses want to improve their AI efficiency but don’t have the resources for it, and it’s not up to the larger corporations to close the discrimination gap. We’re in it for businesses of all sizes; the more efficient we can be with our solutions, the closer we all can be toward fully democratized, responsible AI that works for real people.”
This latest product update continues SolasAI’s focus on innovation and improvement as AI capabilities and importance advance. In June, SolasAI rolled out an integration that tests for algorithmic unfairness using metrics for New York City’s Local Law 144 of 2021 (“NYC 144”). The law requires companies to perform a bias audit on automated employment decision tools and publish these results to ensure fairness for applicants and regulate reliance on AI tools in hiring decisions.
The Bias Explainability & Mitigation Library not only detects the type and scale of algorithmic problems but also explains where the bias comes from, quickly searches for and identifies alternatives with high quality and low disparity, and generates a full report.