To help fight bias in AI, Accenture's Applied Intelligence group developed AI
Fairness Tool. Launched at the CogX conference in June this year, the tool
allows companies and governments to assess whether their AI systems will
treat all people in a fair and unbiased way. Developed in collaboration with the
Alan Turing Institute, the tool's key features include:
DATA ASSESSMENT AND REMOVAL OF RESIDUAL IMPACT: The tool checks
the data that feeds an AI tool and determines what impact sensitive variables
(for example, gender) have on other variables.
AI MODEL ASSESSMENT AND ADJUSTMENT: The tool tests for algorithmic
bias and adjusts the model for equalized impact to make sure people are
treated fairly.
CALCULATION OF ACCURACY TRADE-OFFS: The tool shows whether
improving model fairness will affect its accuracy and to what extent.
Accenture's AI Fairness Tool is part of a suite of offerings to help organizations
create agile, responsible AI that integrates ethical assessments into the
innovation process-without slowing it down. More than ever, organizations
must have strong core values and be ready to defend them vigorously-and AI
needs to reflect them.
The powerful combination of data, analytics and now AI, is fueling socioeconomic growth.
By using the vast amounts of underlying information we
have available and leveraging AI to analyze and harness it, we can
fundamentally improve businesses and the quality of life for people around the
world. We also can deploy AI to address global issues that otherwise look
intractable, like climate change, congestion, pollution, health and many other
problems we face as a planet.
The new imperative of our digital transformation is to balance intelligent
technology and human ingenuity to innovate every facet of business and
become a smarter enterprise.