Income and wealth inequalities have been on the rise globally. According to World Inequality Report 2022, the richest 10% of the global population currently takes 52% of global income, whereas the share of the poorest half is just 8.5% of it. MENA is the most unequal region in the world.
In terms of gender inequality, progress has been very slow at the global level. Women’s share of total labour income increased from 30% in 1990 to just 35% today. Share of women in political representation and in managerial positions at work is still below 30%.
Accelerated use of new technologies brings additional questions to inequality conversation
Because, technological advancements are double-edged swords for existing societal and economic dynamics. They tend to create both winners and losers. While promising for advanced production, transport, healthcare and so on, they also have a role in generating new societal challenges or aggravating existing ones.
Artificial intelligence technologies and their societal implications are worth to discuss in this sense. AI is advancing at pace and seen as a game-changer for every corner of our lives. However, there is a potential that AI powered decisions or automation to produce unintended results, too.
In this article, I would like to address some insights I found critical for us to collectively interpret. Doing so, I hope to raise some questions on principles that can drive AI towards a shared welfare.
First impact area: discrimination and biases against certain groups
AI systems are trained using existing data that inevitably reflects past experiences. Therefore sampling errors or data quality can influence the results of machine learning systems. They may also generate challenges in terms of fairness or bias, even if they trained on accurate volume of data. Unless corrective actions are taken, a person might suffer discrimination, or may not benefit from what technology offers, based on characteristics such as race, age, sex or disability.
To illustrate, applications used to scan candidates for hiring process can be influenced by biases arising from social structures, which are also embedded in dataset. Amazon reportedly experienced this situation in 2018. They found that their recruitment algorithm was biased against women since it works by observing patterns in resumes submitted over a 10-year period, which majority belongs to men candidates.
Another field that algorithmic bias affect is healthcare research and technologies. For instance, a major share of collected heath data belongs to Caucasian patients. Therefore, the Framingham Heart Study cardiovascular risk score is reported to be performed very well for Caucasian patients whereas not delivering accurate results for African American patients.
What about labour market implications?
The power of technology to replicate human labour is nothing new. Think about the first Industrial Revolution. Weaving and spinning machines have played a pivotal role in changing production process. Computers came to scene in the Third and enabled new functions to replace or complement human labour. Today, automation reached to a new level thanks to digitalisation and AI capabilities. Substitution of machines and algorithms for tasks performed by labour is spreading across job categories and industries.
Studies by economists on this challenge, points out to some adverse effects on labour markets and income distributions.
A recent example is the research conducted by Daron Acemoğlu and Pascual Restrapo. Their analysis found that rapid automation was the main driver of changes in the US wage structure over the last four decades. Relative wage declines of workers who specialized in routine tasks are identified closely related with the impact of automation.
Acemoğlu argues that level of automation is excessive:
“When employers make decisions about whether to replace workers with machines, they do not take into account the social disruption caused by the loss of jobs—especially good ones. This creates a bias toward excessive automation.”
Source
Global Development Gap
Another inequality dimension is development gap among countries. IMF researchers found that new technology risks widening the gap between rich and poor countries by shifting more investment to advanced economies where automation is already established. In addition, this trend is expected to increase income inequality, during the transition and possibly in the long run for some groups of workers.
Joseph Stiglitz and Anton Korinek also address this issue in their study. They also point out that certain policy measures including taxation, redistribution and education can help us to mitigate the adverse effects.
Does winner takes all?
AI can act as differentiator among firms by empowering them to respond changes in the market, forecast trends, develop better and cheaper offers comparing to others. Digital platforms are leading the way in utilising a vast amount of data and AI capabilities. They are at the centre of the global economy and affecting all of us individually.
The growing market power of digital platforms raises concerns over unfair market conditions. Strong network effects, large discrepancies in access to data, and aggressive competitive strategies employed by AI are seen as potential threats to free and fair competition. This is why competition authorities are increasingly interested in players in digital markets and taking bold actions to establish new rules for the game.
All these challenges do not mean that we should slow down or hamper AI innovation. Rather, we need to build a bridge between research and practice. AI has potential to help us in climate emergency, better food production, and personalised features for education, increase health and safety measures. To unlock its potential to generate greater shared prosperity, we should look through an impact lens. It is clear that policy and regulations will have an important role for this purpose. Yet, I think business and entrepreneurs also have a responsibility and power to drive social good.
