<<< Start >>>
<<< End >>>
Artificial intelligence (AI): it’s not just about automation. It’s evolving and making new collaboration between humans and machines possible.
This is not a new idea.
J.C.R. Licklider, known as the “father of the internet,” published a proposal in 1960 that outlined the augmentation of human decision-making with intelligent systems.
In effect, he predicted our Fjord 2020 Trend centered on designing intelligence 60 years before we did. But it’s only recently that the data, computing resources and technology have become widely available to help make Licklider’s vision a reality.
When successful, these systems blending human and artificial intelligence help people tackle the increasing complexity and scale of the challenges we face in business and society today.
The view from The Dock
In my role as a data designer at The Dock in Dublin, Accenture’s global R&D and innovation center, we work closely with talented experts across a multitude of disciplines, from data scientists to software engineers to psychologists.
I feel excited and privileged to occupy a unique vantage point where I get to see the real-world impact of the rapid advances in AI-driven technology. We see how it’s helping our clients make great strides in efficiency and accuracy and fostering the creation of a whole new range of products and experiences that were never before possible.
However, as the technology becomes more deeply embedded in the critical systems we rely on, we are also seeing the emergence of new challenges—and even the resurrection of old ones.
Banishing unconscious bias
One such challenge is ensuring algorithmic fairness, which is becoming increasingly important as more decisions of greater importance are made by computer programs.
It is often implied that algorithmic decision-making automatically brings objectivity, but research shows that this is not a given. Unconscious human bias or prejudice can get baked into the algorithms, which can then move on to distribute opportunity or inflict discrimination on people and communities.
Another critical factor to consider with these data-hungry systems is that data always has a history, a lineage and a provenance, all of which offer ample space for latent historical bias to creep in. That bias may, at best, be perpetuated, or at worst, even be amplified by algorithms.
A philosophy of fairness
In the past year, I worked on the Algorithmic Fairness Tool project, a collaboration between Accenture Applied Intelligence, Software Engineering and Fjord.
The aim of this project was to provide users with a means to communicate the complex issues surrounding algorithmic fairness by rendering them more transparent and accessible to a broader audience.
Our design research identified what information was most important for the user and the best way to display it. We also needed to create a tool that could be integrated into the data scientists’ current workflow with minimum disruption or additional effort.
Banking on equitable outcomes
The algorithms and models that decide who can get access to mortgages or loans have a profound, immediate impact on society. Based on a combination of primary and secondary design research, we worked with one of our banking clients to create a high-fidelity prototype that was tested with real data.
What started as a proof of concept as a result of a hackathon between Accenture’s Responsible AI team and academia, was developed into a more practical business tool for examining credit-risk analysis in the banking industry.
We learned that while data science can solve fairness problems from a statistical perspective, some solutions can fall short from an ethical or business perspective. Ensuring more equitable outcomes requires input from the broader organization when it comes to algorithmic decision-making.
By creating tools like the Algorithmic Fairness Tool, we can help to demystify and democratize AI technology. We can also expect to more safely harness the power and vast potential of AI for everyone’s benefit.
Innovate every day and find solutions to the world’s most complex problems. Find your fit with Accenture.
Read my article on Medium for more information on algorithmic equality.
Copyright © 2020 Accenture. All rights reserved. Accenture, its logo, and New Applied Now are trademarks of Accenture.
This document makes descriptive reference to trademarks that may be owned by others. The use of such trademarks herein is not an assertion of ownership of such trademarks by Accenture and is not intended to represent or imply the existence of an association between Accenture and the lawful owners of such trademarks.