Sign up for our newsletter and get the latest big data news and analysis.

Overcoming Bias and Unlocking the Power of Programming Inclusivity

Twentieth-century progress was by and large driven by engineers and physicists. Consider the innovations that changed the complexion of society: highways, urban development, motor vehicles, energy infrastructures, etc.

Consider, as well, how wars were won and lost. In 1942, Werner Heisenberg gathered Germany’s top physicists to discuss the impact nuclear fission — discovered a few years before — might have on the Second World War. The scientists agreed that a nuclear bomb would have the power to end the war. In theory, then, the war could become a race to build such a weapon. Ultimately, the scientists decided it would take too many years to develop and build. Surely the war would be over by then. So they decided not to pursue the project any further.

Three years later, the same physicists, held in captivity by Allied forces, were shocked to discover that the United States had not only developed an atomic bomb but also dropped it on Hiroshima, effectively ending the war. In short, the Allies triumphed because they won a physics race that the Germans chose not to participate in.

In the 21st century, a new race is unfolding. Like before, a line will be drawn between those who can innovate — private and public players alike — and those who don’t compete. This time, it’s not physicists, but coders doing the lion’s share of pushing society forward. But until certain fundamental and underappreciated inequities in informatics are addressed, the entire pace of innovation will be held back. It’s up to nations, corporations, and even individuals to take an inclusive approach to unlock the power of software development.

The challenge of avoiding bias in AI and Big Data

Few politicians receive more attention and scrutiny these days than Alexandria Ocasio-Cortez. AOC recently took some heat for talking about bias when it comes to coding algorithms. Lazy critics were all over her: “Math is impartial and unbiased.”

But, she was right. Coders build code, and they, like every other person on earth, have their own implicit biases. Facial recognition systems, for one, have been at the center of the argument that they are not racially inclusive. A study at the Massachusetts Institute of Technology concluded that facial recognition software displayed an error rate of 34.7 percent for people with dark skin, compared to 0.8 percent for those with light skin. The real world repercussions are far and wide. If programmers, coders, and their departments are cognizant of bias, they are immediately more equipped to overcome it. If coders are not conscious about how they program, the lines of code that they create reflect their biases.

But programming unfair biases don’t rear their ugly head just from the way code is written. Bias is reflected in data inputs. If human-related data does not capture a broad enough sampling of aspects such as geography, education level, and demographics, it will inevitably better serve some people more than others. This is known as the availability heuristic.

In order to avoid this other Big Data biases like confirmation bias, Simpson’s paradox, and non-normality, algorithm creation needs to be transparent. This is the only way to ensure the learning system does not make implicit selections based on the programmer’s bias. Furthermore, the quality of the gathered data would need to be analyzed in order to ensure the proper creation of accurate algorithms free of bias.

The transformative, inclusive power of software and tech development

At January’s World Economic Forum, tech and political leaders discussed the potential of the Fourth Industrial Revolution (4IR) to rid the world of disease, create jobs, and make society run more effectively. But many of the influential people who spoke on the subject made a point to warn of the potential devastating consequences of irresponsible development. “[T]here’s a risk [4IR] will worsen our economic, racial, gender and even our environmental inequalities,” said Marc Benioff, chairman and co-CEO of Salesforce.

In Benioff’s view, access to AI will be a socioeconomic dividing line: “Those without AI will be less educated, weaker, poorer and sicker. So we must ask ourselves, is this the kind of world we want to live in?”

While these questions are just starting to be asked, progress marches onward. In Japan, fully automated AI-powered robots are being tested in train stations, with the end goal of using them in the 2020 Tokyo Olympics. The robot, named “Persuesbot,” uses AI to detect suspicious activity, as well as interpret aggressive body language. Once it detects these anomalies, it sends notifications to the mobile phones of the security staff. These robots have the ability to increase security for major worldwide events, such as the Olympics.

While leaders in the public sector are beginning to form the vocabulary to regulate 4IR, some in the private sector are already showing how it can create a more inclusive world. For example, Augmented Reality (AR) headsets developed by HoloHear are designed to interpret and translate sign language for those with hard-of-hearing. AR also has the ability to show colorblind individuals what colors look like, and enhancing text for people with bad eyesight. Tech development will continue to build towards a world that includes people from all walks of life.

Integrated solutions from both private and public players

Sensible development of AI — both in terms of overcoming bias as well as unlocking its potential to optimize societies and ignite economies — will require collaboration between private and public players. The most feasible way to facilitate this is through legislation. The corporate social responsibilities standards would have to be agreed upon and adopted by these private enterprises.

There’s every reason to believe public and private AI collaboration can create more inclusivity, as these partnerships have proved fruitful in other areas, like sustainability. Take the Houston energy consumption initiative, also known as the WBCSD Energy Efficiency in Buildings 2.0 project (EEB2.0), for example. The goal of this project was to set standards that reduced energy consumption in buildings throughout the city of Houston. To do this, thought leaders, stakeholders, and experts in both the public and private sector got together with the Mayor’s Office of Sustainability. All parties analyzed building value chains, gathering data that would help them create a workable solution.

The result was a coordination platform named  “Energy Efficiency in Buildings – Houston.” Through the platform, which uses a network of self-sustained stakeholders, plans are being created which ensure buildings are made aware of the amount of energy they use. From there, actionable solutions for energy reduction are delivered. With these types of collaborations, we are set to see more integrated solutions that benefit society as a whole.

There’s much to be said about how the race we are running in the 21st century is like that race for a nuclear weapon that would end World War II. But unlike the destruction that this great technological achievement represented, programmers and coders this time around have the opportunity to create innovations that will move us ahead without massive devastation. Through inclusive approaches to development in AI and other technologies, true innovation will take place in a way that’s more efficient and, most importantly, beneficial to all.

About the Author

Mariano Stampella is Business Developer at intive, and one of the seven founding partners of intive-FDV. Computer engineer graduated from the University of Buenos Aires (UBA), he also holds a postgraduate degree in Marketing & Sales for Engineers from the Technological Institute of Buenos Aires (ITBA). As Business Developer of intive, Stampella is responsible for executing the company’s business strategy on the West Coast of the United States. With a strong social commitment, Mariano is one of the founders of Proyecto Nahual, an initiative that promotes social inclusion through training in programming and software testing. In addition, he participates actively in the Commission of Inclusion of the Chamber of Software Companies and Computer Services of Argentina (CESSI).

Sign up for the free insideBIGDATA newsletter.

Leave a Comment

*

Resource Links: