Fairness in 2020.
Discussions around fairness in 2020 will center on both the impact of the pandemic and ongoing societal imbalances, both inside and outside of tech.
With unemployment benefit claims peaking at over 40 million in the U.S., managing return to work will be the biggest challenge of the second half of 2020. Business leaders and politicians will need to work together to ensure this goes as smoothly as possible.
Since many tech companies have been able to continue operations throughout the lockdown period, with some growing remarkably quickly (Amazon hired more than 175,000 additional employees between March and May), resentment may grow if these businesses are not seen to be doing their fair share for society. Arguments over tax bills may be feistier, for example.
The lockdown has also taught us valuable lessons about automating where possible. Autonomous supply chain projects will be accelerated, and we should see exciting developments in autonomous and robotic technology as a result. Questions may be asked, however, if this comes at the expense of employment opportunities for job seekers.
Fairness defined:
Understand your impact on groups and individuals.
Key Trends in Fairness in 2020
FAIRNESS: TREND 1
The Great Displacement: Designing Cobots We Trust
AI in the workplace has a trust problem. A recent Deloitte report found that 63% of 1,000 American executives agreed that “to cut costs, my company wants to automate as many jobs as possible using AI,” and 36 percent believe that AI-related job losses are an ethical issue.
The issues, however, go far beyond job replacement. Algorithmic managers — already a reality for the gig economy — do not allow for human inefficiencies, nor do they lend a listening ear when employees have complaints to make. As more and more automation enters the workplace, companies have an opportunity to design cobots that treat their human colleagues with compassion.
Georgian Impact Podcast
Episode 112: Designing for Humans and Machines with Lindsay Ellerby
“There is a concept in design called seamfulness — it’s allowing the human to see what the machine is really good at and let it do its thing, recognize the differences and sort of celebrate the similarities between humans and machines instead of expecting a machine to be exactly like a human and of course expecting a human to operate at the level of a machine."
Our Prediction
Return-to-Work Protests Evolve
As the negative impacts of COVID lockdowns on employment continue in 2020, the focus of demonstrations will broaden out to include ongoing inequality and job replacement.
Best-in-Class Responses
Provide opportunities for displaced employees to retrain.
AT&T: Retraining the Workforce for Jobs of the Future
Develop internal human computer interaction expertise.
WHAT TO DO
FAIRNESS: TREND 2
A Renewed Focus on Explainable AI in Talent Acquisition
AI has gained a reputation for bias against certain demographics (e.g., women and people of color). Explainable AI, which sheds light on the decision-making of black boxes, is emerging as a key tool to fight this bias.
Our Prediction
Talent Teams Turn to AI
In 2020, we expect to see talent acquisition teams turn to AI to help transform a complex process that can be full of human bias into one that uses data to reach the best outcome for both sides. AI can help to reduce bias, assess cultural fit and improve the candidate experience.
WHAT TO DO
Georgian Impact Podcast
Episode 78: Getting the Bias Out with Cathy O’Neil
“What makes this a problem is that we trust these systems so much … machines, of course, don’t have any prior assumptions, though they can only do so much given the ingredients they’re given as raw data — what they then end up doing is propagating past practices. If we’re talking about a system that decides who deserves a good job, it’ll look like, ‘Who did a good job in the past?’, which is then defined by ‘Who got the job in the past? Who got raises? Who got promotions? Who got fired? Who left early?’ Those people will look like failures, even if it wasn’t them that failed but the culture of the company that failed. A computer’s not going to know the difference between those two things unless we ask it to do more.”
FAIRNESS: TREND 3
Fair and Transparent Pricing
Many software companies moved quickly to offer free products for the duration of the pandemic or to help with COVID-19-specific challenges. With the increased cost of supplies, price increases are tempting but might not be worth it in the long run. In SaaS, price sensitivity is on everyone’s radar for renewals. Customers who have faced headwinds during COVID-19 may struggle to pay in the short term, but the long-term relationship is worth saving.
Our Prediction
Companies Price for Trust
Fair pricing is likely to remain in the headlines as pharmaceutical companies produce vaccines and treatments for the coronavirus. Technology companies that accommodate customers — especially those who faced headwinds — will receive praise and rewards.
Best-in-Class Responses
Proactively work with regulators.
Microsoft's Brad Smith: We Need More Tech Regulation
Define and follow a code of ethics for new technologies.
WHAT TO DO
FAIRNESS: TREND 4
A Spotlight on Procedural and Distributive Fairness in AI
Fairness in AI has emerged as an enormous issue area for policy makers, industry and the public. Individuals may care more about procedural fairness than distributive fairness — that is, people are more willing to accept unfavorable outcomes if they believe the decision-making process is fair. Fairness in ML should be looking both at process-related issues — transparency of the algorithm, ability to voice concern over output, inclusivity of training data, etc. — as well as the distribution of outcomes produced by algorithms.
Georgian Impact Podcast
Episode 93: Facial Recognition, Demographic Analysis and More with Timnit Gebru
"I see the root of the problem being the lack of diversity in this entire field and not just by race and gender, because you can have people of a particular race or gender who still wouldn’t work on this problem or still wouldn’t identify this problem. … At the root of the problem is the concentration of wealth, power and datasets in the companies who are working on AI in general, and what problems they see as being important.”
Our Prediction
Companies Root Out Bias and Hire for Diversity
Tackling bias in AI requires individuals, organizations and government bodies to take a serious look at the source of the problem, which more often than not can be traced to the people creating the AI services in the first place. Countless studies show that the professionals who write AI programs are still largely white males and only 12% of leading machine learning researchers are women.
WHAT TO DO
Fairness: Key Takeaways
When building products, begin by understanding what fairness means to your stakeholders, and design for it. Understand all possible scenarios where unfair outcomes could play out, and develop solutions accordingly.
Businesses have the opportunity to differentiate on fairness by understanding their full impact on all majority and minority stakeholders and designing fair processes and procedures into their systems, products and services. In summary, tech leaders should consider these important points in 2020:
Fairness Checklist
The Great Displacement: Designing Cobots We Trust
A Renewed Focus on Explainable AI in Talent Acquisition
Fair and Transparent Pricing
A Spotlight on Procedural and Distributive Fairness in AI
Continue reading
This post is part of our 2020 State of Trust Report. If you don't see a menu on the left of your screen, dive into the rest of the report here.