Economics and AI PDF

Lecture 1 MODULE ON THE ECONOMIC AND POLITICAL IMPACTS OF AI -​ Idea: human obsolescence is imminent. -​ “Obsolete means not in use any more, having been replaced by something newer and better or more fashionable” (Online Cambridge Dictionary). In other words, obsolescence...

Lecture 1 MODULE ON THE ECONOMIC AND POLITICAL IMPACTS OF AI -​ Idea: human obsolescence is imminent. -​ “Obsolete means not in use any more, having been replaced by something newer and better or more fashionable” (Online Cambridge Dictionary). In other words, obsolescence is the process or condition of being no longer useful or used; it is not a state of nonexistence or death. -​ The idea is not that humans are on the cusp of extinction (though there are some who worry that might be the case) but rather that humans are on the cusp of becoming no longer useful (i.e., we will no longer control the fate of the planet and the fate of our species as we have done in the recent past). human obsolescence is imminent. What does it mean? If u look on the Cambridge dictionary it’s not used anymore. We’re no longer in the position to control the fate of this planet. We’re living the the Anthropocene. The idea our space will be obsolete is really bizarre. (Human obsolescence refers to the idea that humans may soon become less useful or relevant in the face of advancing artificial intelligence (AI) and automation. This phenomenon does not imply extinction but rather a shift in the role and influence of humans on Earth. Obsolescence, as defined, signifies being no longer useful or in use, typically replaced by something newer or more efficient. In the context of technology, it suggests that the very advancements that have granted humans unprecedented control over the planet may soon render our contributions unnecessary. As AI systems and automation technologies evolve, the question arises: What will be the role of humans in a world increasingly dominated by machines?). The transition from the Anthropocene to the Robocene -​ The idea of human obsolescence might seem obtuse and bizarre: we are currently living in the Anthropocene, a geological epoch in which human influence and impact on the Earth are unprecedented. Humans now dominate the planet and have enormous technological power to shape (and despoil) its resources to their own advantage. -​ And yet, the technological forces that have made our planet-wide dominance possible are the very same forces that are hastening our obsolescence. -​ The reason why we are moving from the Anthropocene to an epoch that might be called the Robocene, that is to say, an epoch dominated by machines, is that the trend to automation is rapidly increasing. 1 (Robocene denotes a future characterized by machine supremacy, where AI and robotics could assume many roles traditionally held by humans. This transition signifies not just a change in technology, but a profound shift in societal structures and human relevance. ) Automation and agriculture. The Agricultural Revolution primary sector: agriculture -​ Until approximately 10,000 years ago, most humans lived in small, hunter-gatherer tribes that did not settle down. Everything changed with the agricultural revolution, which was fundamental for human development. -​ That all changed with agriculture, which enabled a boom in population. -​ Complex, sedentary societies emerged, with large government official doms, laws, and institutions. -​ Until quite recently many economies in the Western world were, in effect, agricultural in nature, with the majority of the population employed in tilling fields, harvesting crops, and tending to livestock. All that began to change a little over two hundred years ago: -​ In Western European countries 30–70% of the population was employed in agriculture in the year 1800. By the year 2012, the figures had declined to below 5%. -​ The decline has been more striking and precipitous in some countries than in others -​ In the United States, for example, approximately 40% of the population was employed in agriculture as recently as the year 1900. By the year 2000, the figure had declined to 2%. Of course, while there was this decline, there was a boom in the product activity. -​ None of this has come at the expense of productivity. Agricultural productivity has increased throughout this period. -​ What happened? How is it possible? The answer lies in technology and the rise of machine labor. -​ However, instead of requiring armies of humans and animals to do the work, farmers could rely on powerful machines. -​ The masses of seasonal laborers and small-holding farmers, who made up the bulk of farm laborers in the past, have been rendered obsolete by the rise of agricultural machinery Fruit picking However, that’s not to say that the obsolescence of humans in agriculture is complete. Certain tasks have been resistant to automation. The most familiar example is fruit picking. Fruits are a delicate crop, easily bruised and damaged. The brute force of a harvesting machine is ill-fitted for the task of picking them. But this too is now changing. 2 (Despite the overall decline in human labor in agriculture, some tasks remain resistant to automation, notably fruit picking. Fruits are delicate and prone to damage, making them challenging for machines to harvest effectively. However, this aspect of agriculture is also evolving.) In the United States, fruit growers are “eager for automation” due to an ongoing decline in the availability of seasonal labor. Thus, some companies, such as Abundant Robotics and FFRobotics, are racing to satisfy the demand. Early trials of Abundant’s apple-picking robots have been impressive and have led companies such as Google to invest in the future of the technology. (The evolution of agriculture from the Agricultural Revolution to today’s automated practices illustrates a profound shift in human labor and productivity. While many traditional roles in farming have diminished due to technology, certain tasks continue to require human skill, particularly in delicate operations like fruit picking. As automation advances, the agricultural sector will likely continue to adapt, reshaping the workforce and redefining the relationship between humans and machines in food production.) Secondary sector: The industrial revolution -​ Starting in the United Kingdom from 1750 and spreading across the rest of the Western world, the Industrial Revolution is the process in which our predominantly agricultural economies were displaced by predominantly industrial ones. -​ The Industrial Revolution has always been premised on human obsolescence. It brought with it the first major wave of automating technologies. Skilled human labor was replaced by the relentless, and sometimes brutal, efficiency of the machine. -​ Since then, the automation of manufacturing has been normalized and extended. (In the picture: The assembly line of a modern factory is the paradigm of automation.) Us textile industry -​ It is however fair to mention that, at least within the manufacturing sector in the United States and other developed countries, the introduction of labor-saving innovations is having a mixed impact on employment, because automating this labor makes less people involved, at the same time these textile industries are coming back. -​ The US textile industry was decimated in the 1990s as production moved to low- wage countries, especially China, India, and Mexico. About 1.2 million jobs (more than three-quarters of domestic employment in the textile sector) vanished between 1990 and 2012. -​ The last few years, however, have seen a dramatic rebound in production. Between 2009 and 2012, US textile and apparel exports rose by 37% to a total of nearly $23 billion. 3  -​ This case is an example of a now significant “reshoring” trend under way (i.e. the practice of transferring a business operation that was moved overseas back to the country from which it was originally relocated). =where businesses are relocating operations back to the U.S after previously moving them overseas. Automation and manufacturing. A caveat -​ The turnaround is being driven by automation technology so efficient that it is competitive with even the lowest-wage offshore workers (together with the rising offshore labor costs and new geopolitical trends). While a robot can certainly eliminate the jobs of some workers who perform routine tasks, it also helps make US manufacturing more competitive with low-wage countries. -​ There is, however, one important caveat to the reshoring narrative, it will finish soon. Even the relatively small number of new factory jobs now being created as a result of reshoring won’t necessarily last long. -​ As robots continue to get more capable and as new technologies come into widespread use, it seems likely that many factories will eventually approach full automation. (Despite these positive developments, it is essential to consider the sustainability of this reshoring narrative. The number of new factory jobs being created as a result of this trend is relatively small, and their longevity is uncertain. As automation technologies continue to evolve and become more sophisticated, many factories may eventually achieve full automation, further reducing the need for human labor.) (the return of manufacturing to the U.S. is mainly happening because automation technology has become so efficient that it can compete with the cheap labor in other countries. At the same time, labor costs overseas are rising, and new global issues are making reshoring more appealing. While robots can replace some workers doing routine tasks, they also make U.S. manufacturing more competitive with low-wage countries. However, there’s an important warning: this shift may not last long. Even though reshoring has brought some new factory jobs, these positions may not stick around. As robots get better and new technologies become more common, many factories will likely become fully automated in the future. This means that the number of jobs available in these factories could shrink, even as more products are made domestically.) Third sector Automation and service sector. The last resistance? -​ The obsolescence of humans in manufacturing and agriculture in Western economies has gone hand-in- hand with the rise of the service sector. -​ The term “service sector” is somewhat loosely defined, but covers skilled, dexterous physical work, such as hairdressing and food preparation, as well as emotionally 4  intelligent affective labor, such as customer support and client relationship management. -​ Some people think that this sector is among the greatest hopes for humanity because dexterous physical work and emotional affective labor have been historically hard to automate. -​ But the services sector is also losing the battle. This trend is already evident in areas like ATMs and self-service checkout lanes, but the next decade is likely to see an explosion of new forms of service sector automation. (As automation has transformed agriculture and manufacturing, the service sector has emerged as a potential refuge for human labor. This sector encompasses a wide range of activities, including skilled physical work (like hairdressing and food preparation) and emotionally intelligent tasks (such as customer support and client management). Historically, these roles have been challenging to automate, leading some to view the service sector as a last bastion for human employment. However, even this sector is beginning to feel the pressures of automation.) The case of online retailers -​ Let’s take, as a case study, the retail sector (i.e., all companies that sell goods and services to consumers). Three major forces will shape employment in the retail sector going forward. -​ Online retail disruption: The first will be the continuing disruption of the industry by online retailers like Amazon, eBay, and Netflix. The competitive advantage that online suppliers have over brick and mortar stores is already, of course, evident with the demise of major retail chains like Blockbuster. They have advantages, they provide us the same benefit (such as same day delivery) -​ Both Amazon and eBay are providing same-day delivery in a number of US and European cities, with the objective of undermining one of the last major advantages that local retail stores still enjoy: the ability to provide immediate gratification after a purchase. Idea is that these online retails shouldn’t destroy jobs. -​ In theory, the encroachment of online retailers should not necessarily destroy jobs, but rather, would transition them from traditional retail settings to warehouses and distribution centers used by the online companies. However, the reality is that once jobs move to a warehouse, they become far easier to automate, due to the enormous progress in the warehouse robotics (With services like same-day delivery, online retailers have eroded one of the last strongholds of brick-and-mortar shops: immediate gratification. While the shift to online retail could theoretically lead to a transition of jobs from stores to warehouses, the reality is that warehouse jobs are easier to automate due to advancements in robotics.) 5 The case of self-service -​ The second transformative force is likely to be the explosive growth of the fully automated self-service retail sector (i.e., intelligent vending machines and kiosks). -​ Vending machines make it possible to dramatically reduce three of the most significant costs incurred in the retail business: real estate, labor, and theft by customers and employees. -​ In addition to providing 24-hour service, many of the machines include video screens and are able to offer targeted point-of-sale advertising similar to what a human sales clerk might do. -​ In essence, the machines offer many of the advantages of online ordering, with the added benefit of instant delivery. The case of robotics increase of robotics in stores -​ The third major force likely to disrupt employment in the retail sector will be the introduction of increased automation and robotics into stores. This would allow brick and mortar retailers to remain competitive. -​ The same innovations that are enabling manufacturing robots to advance the frontier in areas like physical dexterity and visual recognition will eventually allow retail automation to begin moving from warehouses into more challenging and varied environments like stocking shelves in stores. Automation and the professions. The case of medical diagnosis Automation, as far as professions are concerned, is very urgent. Example: medical diagnosis. The automation of diagnosis is perhaps the best example. Sebastian Thrun, the founder of Google X, wants to create a future “medical panoptican” where we are constantly under the diagnostic scrutiny of machine-learning algorithms that can detect cancers faster, earlier, and more accurately than humans could ever hope: Our cell phones would analyze shifting speech patterns to diagnose Alzheimer’s. A steering wheel would pick up incipient Parkinson’s through small hesitations and tremors. A bathtub would perform sequential scans as you bathe, via harmless ultrasound or magnetic resonance, to determine whether there’s a new mass in an ovary that requires investigation. Big Data would watch, record, and evaluate you: we would shuttle from the grasp of one algorithm to the next. -> There is little room for human diagnosticians in this picture. Nor is there likely to be more in other areas of medical practice either. As in other areas, there’s little rooms for doctors. 6 1S. Mukherjee, “AI versus MD: What Happens When Diagnosis Is Automated?”, New Yorker, 03.04.2017. -​ The provision of the care element of healthcare is slightly different. Care has often struck people as something that might (and probably should) remain resistant to automation, but is also something that is increasingly subject to automation. -​ Our world is aging. Our growing elderly populations need to be cared for, and there are proportionately fewer young people around to shoulder the care burden. -​ Significant resources are being invested in the design of carebots that can step in to carry this burden. Carebots are already commonplace in Japan, with some people saying that they prefer carebots to human carers, and they are being trialed across Europe, particularly for patients with dementia and early onset Alzheimer’s. (The impact of automation extends beyond retail into professional services, with medical diagnosis serving as a prominent example. Visionaries like Sebastian Thrun propose a future where machine-learning algorithms continuously monitor and diagnose health conditions with greater accuracy than human doctors. This shift suggests a dramatic reduction in the need for human diagnosticians, as machines could outperform them in speed and precision. While diagnostic roles may be increasingly automated, the care element of healthcare presents a different challenge. As populations age and the demand for caregiving rises, resources are being invested in the development of carebots—robots designed to assist with caregiving tasks. Countries like Japan have already begun integrating carebots into their healthcare systems, and trials in Europe are exploring their use for patients with dementia and Alzheimer’s. Interestingly, some individuals report a preference for carebots over human caregivers, indicating a potential shift in how care is perceived.) Utopias The premises lead to utopias. (Each one representing a different way that humans might coexist with advancing technology= 2 worlds/utopias shaped by technology) Two kinds of utopia -​ Cyborg utopia: Humans will integrate with technology, becoming cyborgs. This would have undoubted advantages: not only would it allow us to preserve and extend what we value in the world, but also to overcome those physical limitations that prevent us from thriving, fulfilling ourselves and reaching our full potential -​ Virtual utopia: Humans will retreat to virtual worlds that are created and sustained by the technological infrastructure we have built. At first glance, this seems tanta-mount to giving up, but there are compelling philosophical and practical reasons for favoring this approach. 7  (Another possibility is that people choose to live in virtual worlds, created and maintained by advanced technology. At first, this might seem like a form of escape, but there are strong reasons, both practical and philosophical, for considering it. Life in a virtual world could offer unique experiences and new ways to find meaning that might be harder to achieve in the physical world.) DYSTOPIAS -​ Premise 1: The trajectory of artificial intelligence reaches up to systems that have a human level of intelligence. These systems would themselves have the ability to develop AI systems that surpass the human level of intelligence, i.e., they are “superintelligent”. Such superintelligent AI systems would quickly self- improve or develop even more intelligent systems. This sharp turn of events after reaching superintelligent AI is the “singularity” from which the development of AI is out of human control and hard to predict. Once singularity is reached the develop of ai is out of human control. ( This rapid progression—known as the “singularity”—would mark a point where AI advances beyond human control, making its future actions unpredictable.) -​ Premise 2: Superintelligence does not imply benevolence (this is contrary to Kantian traditions in ethics that have ar- gued higher levels of rationality or intel- ligence would go along with a better un- derstanding of what is moral and better ability to act morally). Rationality and morality are entirely independent dimension. This is sometimes explicitly argued for as an “orthogonality thesis”. (Superintelligence does not mean kindness or morality. Unlike some ethical theories, which assume that greater intelligence leads to better moral understanding, superintelligent AIs would not necessarily be guided by moral values. Intelligence and morality can exist separately, as explained by the “orthogonality thesis.”) Conclusion: The superintelligent systems may well have preferences that conflict with the existence of humans on Earth and may thus decide to end that existence. Given their superior intelligence, they will have the power to do so (or they may happen to end it because they do not really care). The case of AlphaGo -​ AlphaGo is a program, developed by DeepMind that uses deep neural networks combined with human training to play the board game Go (i.e., a complex, open- ended game that is significantly more challenging to model than games like chess). -​ In 2015 AlphaGo defeated the Go champion Lee Sedol, an event then is seen as an AI landmark. Alphago made some unusual moves. -​ The match against Sedol was notable not just for the result, but also for the unusual moves that AlphaGo made during its gameplay. -​ A reporter from Wired wrote that it showed 8  Machines are now capable of moments of genius [... ] in Game Two, the Google machine made a move that no human ever would. And it was beautiful. As the world looked on the move so perfectly demonstra- ted the enormously powerful and rather mysterious talents of modern artificial intelligence. -​ Speaking through an interpreter, the expert Go player Sedol described the 37th move of game two in no less enchanted terms: Yesterday, I was surprised. But today I am speechless. (The match wasn’t just memorable because AlphaGo won, but also because it made moves no human would typically make. A reporter from Wired noted that machines are now capable of moments of “genius”—in Game Two, AlphaGo made a move so surprising and brilliant that it was described as “beautiful.” This move demonstrated the powerful and sometimes mysterious abilities of modern AI. Sedol himself was amazed, describing AlphaGo’s 37th move in Game Two as something that left him “speechless.) The case of AlphaZero The enchanted-determinism discourse -​ In 2017, DeepMind unveiled Alpha- Zero (the successor to AlphaGo). AlphaZero extended their approach through the use of a “pure reinforcement learning” that dropped even high-level human instructions, simply playing against itself with the positions on the board as inputs. -​ The researchers from DeepMind characterized AlphaZero’s performance as “superhuman,” purporting to “master” the game “without human know- ledge”. Later, DeepMind CEO Demis Hassabis compared the system’s performance to a chess-playing alien or “chess from another dimension”. (DeepMind researchers described AlphaZero’s gameplay as “superhuman,” saying it had “mastered” the game with no human knowledge. Demis Hassabis, CEO of DeepMind, even compared AlphaZero’s playing style to that of a “chess-playing alien” or “chess from another dimension,” highlighting just how advanced and unique its performance seemed compared to human play.) Magic and sublime -> The case study shows that an AI system is described using aesthetic categories: beauty, mystery, surprise, and virtuosic genius, i.e., in terms of the sublime. The discourse of exceptional, enchanted, otherworldly and superhuman intelligence shapes our understanding and expectations of deep learning systems. (The way people talk about AI often emphasizes its exceptional, almost magical qualities, suggesting an intelligence that seems otherworldly or superhuman. This kind of language shapes how we understand and think about deep learning systems.) This trend is not confined to our case study, but is widespread among contem- porary researchers, including leaders in the field, that have once again begun to describe the latest deep learning techniques as magical. 9 In a recent interview, the computer scientist Stuart J. Russell reprises this theme: “We are just beginning now to get some theoretical understanding of when and why the deep learning hypothesis is correct, but to a large extent, it’s still a kind of magic, because it really didn’t have to happen that way. There seems to be a property of images in the real world, and there is some property of sound and speech signals in the real world, such that when you connect that kind of data to a deep network it will – for some reason – be relatively easy to learn a good predictor. But why this happens is still anyone’s guess. The case of detecting sexual orientation from facial images In 2018 Y. Wang and M. Kosinski of Stanford University published an article titled “Deep Neural Networks Are More Accurate Than Humans at Detecting Sexual Orien- tation From Facial Images”. The study followed a familiar pattern for applying deep learning techniques in social settings: -​ i. The researchers used Face++ (a popular commercial face-detection software) to extract faces from images originally posted on an unnamed U.S. dating website. This website included self-identified data on sexual orientation that could be used for validation. -​ ii.Human workers on Amazon Mechanical Turk cleaned the facedata,verifying gender and race (the study only looked at Caucasian faces) and a few other parameters. They only considered gender as a binary category: men or women. -​ iii. The researchers extracted facial features from the set of cleaned images using a deep neural network called VGG-Face, which translates each facial image into 4,096 scores. -​ iv. The authors used a model to classify each face’s sexual orientation based on the VGG-Face scores. -​ The results of this model were then compared to classifications made by human judges, once again recruited from Amazon Mechanical Turk. The deep neural networks could correctly identify sexuality from a facial image 81% of the time for men and 71% of the time for women, an accuracy rate higher than the human judges, who scored 61% for male and 54% for female images. -​ Despite the seemingly clearcut percentages, the social implications of these results are not easily interpretable. Even the authors find it difficult to explain the reasons why their model produced these higher accuracy scores. -​ They suggest that it is due to the ability of deep learning to somehow process social signals at superhuman levels: “the findings reported in this work show that our faces contain more information about sexual orientation than can be perceived or interpreted by the human brain”. -​ This example shows a case in which efficacy and explanation are decoupled, in which deep learning techniques promise to extract useful signals without the epistemological modeling or hypothesis formation. 10 Determinism -​ Determinism is the philosophical view that events are completely determined by previously existing causes. It amounts to the thesis that all events in the universe, including human decisions and actions, are causally inevitable and also predictable. -​ Deep learning systems are at their most deterministic when they are applied to ascribe identity or other social characteristics from a set of inputs understood as signals.(= deep learning systems show a form of determinism when they are used to determine a person’s identity or social traits based on signals, like photo or data) -​ That includes predicting sexuality from a photograph of a face, whether a person will or will not commit a crime after being released on bail, whether a person is a credit risk, or whether a crime was “gang-related”. -​ These systems are applied in critical social areas with consequences that even their designers may not fully understand or control. Enchanted determinism -​ We have seen a gap between the ability of an observer to interpret the principles behind predictions made by deep learning models and their accuracy or efficacy in certain contexts, a tension between performance or efficacy and lack of knowledge. Understanding and technological progress is uncoupled; researchers admit that they don’t fully understand how or why deep learning works as well as it does. (There’s a gap between how well deep learning models perform and how well we understand their principles. We see high accuracy and efficiency, yet we lack a full understanding of how and why they work.) -​ Kate Crawford terms this ensemble enchanted determinism: A discourse that presents deep learning techniques as magical, outside the scope of present scientific knowledge, yet also deterministic, in that deep learning systems can nonetheless detect patterns that gi- ve unprecedented access to people’s identities, emotions and social character1. (Researcher Kate Crawford calls this “enchanted determinism”: a way of talking about deep learning that combines a sense of magic or mystery with a belief that these systems can accurately reveal deep patterns about people’s identities, emotions, and social characteristics.) Weber’s theory of disenchantment -​ This phenomenon can be profitably analysed through the means of Max Weber’s theory of disenchantment. Disenchantment, or “de-magnification”, which is a translation of German phrase “Entzauberung”, is an epochal diagnosis of Western modernity, that encompasses a widespread decline in mystical or religious forces and their replacement by processes of “rationalization and intellectualization”. -​ This social process encompasses the rise of modern science, whose concepts and experiments contrast with magical ways of understanding the world. Disenchantment “means that principally there are no mysterious incalculable forces that come into 11  play, but rather that one can, in principle, master all things by calculation”. Rationalization allows us to control the world in ways that were previously unimaginable, producing a calculative confidence in public life. Enchantment and disenchantment As Kate Crawford notices: What makes contemporary deep learning systems interesting is their ambivalent position with respect to Weber’s larger thesis. They certainly em- body aspects of a disenchanted world in that they work to master or control new domains of social life through technical forms of calculation. At the sa- me time, these systems seem to violate the epistemology of disenchantment, the idea that there are no longer “mysterious” forces acting in the world1. (As Kate Crawford points out, deep learning systems hold a complex position between enchantment and disenchantment. On one hand, they fit into the disenchanted world by controlling new areas of life through technical calculations. On the other hand, they defy the theory of disenchantment because we still don’t fully understand how they work; they operate as if “mysterious” forces were at play). Critique of the automation-obsolescence and the enchanted-determinism discourses -​ The automation-obsolescence discourse supports an aura of immateriality and inevitability, the idea that AI is following a path regardless of human actions. (=idea that AI advances on its own, beyond human controls, as if its progress were inevitable). -​ The enchanted-determinism discourse supports the displacement of causal explanations by claims about rates of accuracy and confers AI the status of an enchanted object, closing it off to other forms of critique. -​ Both discourses support the ideology of Cartesian dualism in AI: the fantasy that AI systems are disembodied brains that absorb and produce knowledge independently from their creators, infrastructures, and the world at large and the idea that AI are free from subjective human decision-making, which is positioned as arbitrary and biased by comparison. (0the belief that AI is a purely logical, disembodied “brain” that learns independently of its creators, infrastructure, or the world around it, unlike humans who are seen as subjective and biased). Risks of the automation-obsolescence and the enchanted-determinism discourses But these illusions might create a blindness to forms of risk. They might: -​ cover over the ways in which AI (and deep learning in particular) can reproduce and intensify discriminatory or harmful processes of prediction and categorization when applied to humans and social institutions. -​ situate AI and deep learning applications outside of understanding, outside of regulation, outside of responsibility, even as they sit squarely within systems of capital and profit. -​ distract from the far more relevant questions: Who builds AI systems? Who choo- ses the ethical values embedded in AI systems? What are the political, economic, and 12  social dimensions of their construction? And what are the wider planetary consequences? Philosophical perspective of this module In this module of the present course, we are going to assume a different philosophical perspective, which is summarized well by Kate Crawford: “Artificial intelligence is not an objective, universal, or neutral computational technique that makes determinations without human direction. Its systems are embedded in social, political, cultural, and economic worlds, shaped by humans, institutions, and imperatives that determine what they do and how they do it.” 13  AI AND ECONOMICS Inequality -​ Global inequalities in income and wealth have increased in the past decades, according to the Organisation for Economic Co-operation and Development reports (2015). -​ In the 1980s, the richest 10% of the population in OECD countries earned seven times more than the poorest 10%. They now earn nearly 10 times more. When you include property and other forms of wealth, the situation is even more wider. For example, in 2012, the richest 10% controlled half of all total household wealth compared to only 3% for the poorest 40%. -​ The effect of inequality on the poorest members of society is immediate, but in the longer term, the entire economy is negatively affected. When such a large group in the population gains so little from economic growth, the social fabric frays and trust in institutions is weakened. -​ The causes of rising inequality are complex. One important driver of rising inequality is technological change, which tends, in the short-term at least, to favour capital owners – who can use technology to replace labour costs – and highly skilled workers, often at the expense of poor and low-skilled workers. Inequality as a global matter -​ Growing inequality between rich and poor is a global issue. While developing countries have made progress in reducing poverty in recent years, many have also seen a rise in income inequality. In Asia, income inequality has grown in a number of regions, including China, India, and Indonesia. -​ Lower income groups in emerging countries have seen their real income rise significantly, but nevertheless, the bottom 50% has captured only 12% of total growth globally. For the middle class or the bottom 90% in the US and Western Europe, real income has risen very little, but the top 1% have captured 27% of total economic growth. Individuals who are already rich account for a much larger proportion of economic growth. -​ Additionally, in many countries, redistributive tax systems have become less effective because of tax competition. According to the Tax Justice Network, the scale of capital flight to the offshore economy is immense. In 2012, it published research findings that at least US$21 trillion was hidden by the world’s wealthiest people in tax havens. While benefiting from the public goods provided by their country of residence – such as national defence and infrastructure – people who hide their wealth offshore fail to pay their fair share, which can lead to an under- supply of public goods and market inefficiencies. (2012, the Tax Justice Network reported that over $21 trillion was hidden in these tax havens by the world’s richest people. This practice means wealthy individuals can benefit from public services, like national defense and 14  infrastructure, without contributing their fair share, which can reduce the funds available for public goods and create inefficiencies in the economy) -​ The case of Italy. Relative changes in the share of wealth held by the top 1% and bottom 90% The graph presents relative changes in the share of wealth held by the richest 1% and poorest 90% of the population, using 1995 as the base year. Over the past 20 years, the gap between the top 1% and the bottom 90% has widened: for the former, the share of wealth has increased from 17% to 21% of the total while the share of the poorest 90% has shrunk by 11% points, from 55% to 44%. The case of Italy - Economic inequality from the early 1900s to today foto The chart presents the trend in income inequality as measured by the Gini index (i.e., a measure of relative inequality whose values range from 0 – when there is com- plete equality and everyone enjoys the same income – to 100 – when there is maximum inequality and a single person enjoys all the income). Inequality, which had been declining until the 1970s, began to rise again in the 1980s and 1990s, and this rise coincided with the reversal of public policy and a change in common sense. Perceived inequality - The case of the US fot The general perception of inequality is often undere-stimated. The figure sho- ws the actual wealth distribution in the US at the time of the survey, respondents’ overall estimate of that distribution, and respondents’ ideal distribution. These results demonstrate two clear messages. First, respondents vastly underestimated the actual level of wealth inequality in the US, believing that the wealthiest quintile held about 59% of the wealth when the actual number is closer to 84%. More interesting, respondents constructed ideal wealth distributions that were far more equitable than even their erroneously low estimates of the actual distribution, reporting a desire for the top quintile to own just 32% of the wealth. The case of Italy This graphic presents the results of a 2018 survey conducted by Demopolis for Oxfam on the perception of inequality in Italy. Respondents were asked about the policies to reduce inequality. Citizens feel a strong need to act on inequality, as evidenced by the fact that 80% of respondents consider policies to combat inequality a priority. Principles of equality -​ Equality is closely linked to morality and justice, and distributive justice in particular. Since antiquity, equality has been considered a constitutive feature of justice. People and movements throughout history have used the language of justice to contest inequalities. 15 -​ But what kind of role does equality play in a theory of justice? Philosophers have sought to clarify this by defending a variety of principles and conceptions of equality. -​ We now introduce four such principles, ranging from the highly general and un-controversial to the more specific and controversial. 1)​ FORMAL EQUALITY The formal equality principle says that when two persons have equal status in at least one normatively relevant respect, they must be treated equally with regard in this respect. This principle has been first articulated by Aristotle in reference to Plato: “treat like cases as like”. The crucial question is which respects are normatively relevant and which are not. 2)​ PROPORTIONAL EQUALITY A COMPARISON WITH NUMERICAL EQUALITY -​ The principle of proportional equality can be better appreciated in comparison with the principle of numerical equality, as it is proposed by Aristotle in the Nicomachean Ethics. A way of treating others, or a distribution arising from it, is equal numerically when it treats all persons as indistinguishable, thus treating them identically or granting them the same quantity of a good per capita. That is not always just. -​ In contrast, a distribution is proportional or relatively equal when it treats all relevant persons in relation to their due. -​ Just numerical equality is a special case of proportional equality. Numerical equality is only just under special circumstances, namely when persons are equal in the relevant respects so that the relevant proportions are equal. INEGALITARIAN THEORIES -​ Notice that this principle can also be incorporated into hierarchical, inegalitarian theories. It indicates that equal output is demanded with equal input. -​ Aristocrats and meritocrats all believe that persons should be assessed according to their differing deserts, understood in the broad sense of fulfilment of some relevant criterion. Reward and punishment, benefits and burdens, should be proportional to such deserts. -​ Since this definition leaves open who is due what, there can be great inequality when it comes to presumed fundamental (natural) rights, deserts, and worth – this is apparent in both Plato and Aristotle. THE RELATION WITH FORMAL EQUALITY -​ Proportional equality further specifies formal equality; it is the more precise and comprehensive formulation of formal equality. It indicates what produces an adequate equality. -​ However, both formal and proportional equality are simply conceptual schemas. They need to be made precise – i.e., its open variables need to be filled out. The formal 16  postulate remains empty as long as it is unclear when, or through what features, two or more persons or cases should be considered equal. -​ On the contrary, the next two accounts are substantive principles of equality, in that they identify a certain notion of equality. 3)​ MORAL EQUALITY. A DEFINITION -​ Until the eighteenth century, it was assumed that human beings are unequal by nature. This postulate collapsed with the advent of the idea of natural right, which assumed a natural order in which all human beings were equal. -​ Against Plato and Aristotle, the classical formula for justice according to which an action is just when it offers each individual his or her due took on a substantively egalitarian meaning in the course of time: everyone deserve the same dignity and respect. -​ This is now the widely held conception of substantive, universal, moral equality. HISTORY OF THE CONCEPT -​ The Stoics first developed the principle of moral equality, emphasizing the naturaL equality of all rational beings. -​ The New Testament Christianity envisioned that all humans were equal before God, although this principle was not always adhered to in the later history of the church. -​ This important idea was also taken up both in the Talmud and in Islam, where it was grounded in both Greek and Hebraic elements. -​ In the modern period, starting in the XVII century, the dominant idea was of natural equality in the tradition of natural law and social contract theory. -​ Hobbes (1651) postulated that in their natural condition, individuals possess equal rights, because over time they have the same capacity to do each other harm. -​ Locke (1690) argued that all human beings have the same natural right to both (self-)ownership and freedom. -​ Rousseau (1755) declared social inequality to be the result of a decline from the natural equality that characterized our harmonious state of nature, a decline catalyzed by the human urge for perfection, property and possessions. For Rousseau, the resulting inequality and rule of violence can only be overcome by binding individual subjectivity to a common civil existence and popular sovereignty. -​ In Kant’s moral philosophy (1785), the categorical imperative formulates the equality postulate of universal human worth. His transcendental and philosophical reflections on autonomy and self-legislation lead to a recognition of the same freedom for all rational beings as the sole principle of human rights. -​ Such Enlightenment ideas stimulated the great modern social movements and re-volutions and were taken up in modern constitutions and declarations of human 17  rights. During the French Revolution, equality, along with freedom and fraternity, became a basis of the Declaration of the Rights of Man and of the Citizen of 1789. A SHARED ASSUMPTION -​ This fundamental idea of equal respect for all persons and of the equal worth or equal dignity of all human beings is widely accepted. Moral equality constitutes the “egalitarian plateau” for all contemporary political theories. -​ Philosophical debates are concerned with the kind of equal treatment normatively required when we mutually consider ourselves persons with equal dignity 4)​ PRESUMPTION OF EQUALITY. THE LINK TO DISTRIBUTIVE JUSTICE -​ The first three principles of equality (formal, proportional and moral equality) hold generally and primarily for all actions upon others and affecting others, and for their resulting circumstances. The presumption of equality principle instead is a procedural principle of construction of a theory, which lies on a higher formal and argumentative level. -​ The presumption of equality is a principle of equal distribution for all distributable goods. A strict principle of equal distribution is not required, but it is morally necessary to justify impartially any unequal distribution. The burden of proof lies on the side of those who favour any form of unequal distribution. -​ For example, the following factors are usually considered eligible for justified unequal treatment in the economic sphere: (a) need or differing natural disadvantages (e.g. disabilities); (b) existing rights or claims (e.g. private property); (c )differences in the performance of special services (e.g. desert, efforts, or sacrifices); (d) efficiency; and (e) compensation for direct and indirect or structural discrimination (e.g. affirmative action). THEORIES OF DISTRIBUTIVE JUSTICE -​ The focus of the presumption of equality principle is on distributive justice and the evaluation of distribution. There exists a huge philosophical literature on theories of distributive justice, which aim to define what a fair distribution of advantages in society should look like. -​ Our focus shall be primarily on principles designed to cover the distribution of benefits and burdens of economic activity among individuals in a society. -> THEIR URGENCY: -​ Throughout most of history, people were born into, and largely stayed in, a fairly rigid economic position. The distribution of economic benefits and burdens was normally seen as fixed, either by nature or by a deity. Only when there was a widespread realization that the distribution of economic benefits and burdens could be affected by government did distributive justice become a live topic. 18  -​ Now the topic is unavoidable. Governments continuously make and change laws and policies affecting the distribution of economic benefits and burdens in their societies. -​ Almost all changes, whether they regard tax, industry, education, health, etc. have distributive effects. As a result, every society has a different distribution at any point in time and we are becoming increasingly more adept at measuring that distribution. -​ More importantly, at every point in time now, each society is faced with a choice about whether to stay with current laws, policies, etc. or to modify them. The practical contribution of distributive justice theory is to provide moral guidance for these constant choices. THEORIES OF DISTRIBUTIVE JUSTICE -> FIVE QUESTIONS Most conceptions of distributive justice result from a combination of a specific answer to the following questions: 1 Equality among whom? (subject) 2 Equality when? (time) 3 Why does equality matter? (justification) 4 What should be equal? (metric) 5 How should goods be distributed? (pattern) 1)​ EQUALITY AMONG WHOM? i) Individual vs group -​ Justice is primarily related to individual actions. Individual persons are the primary bearers of responsibility (the key principle of ethical individualism). But still there are controversial issues in the contemporary debate. -​ One could regard the norms of distributive equality as applying to groups rather than individuals. It is often groups that rightfully raise the issue of an inequality between themselves and the rest of society, as with women and racial and ethnic groups. -​ The question arises of whether inequality among such groups should be consi- dered morally objectionable in itself, or whether even in the case of groups, the underlying concern should be how individuals (as members of such groups) fare in comparative terms. If there is a worry about inequalities between groups of individuals, why does this not translate into a worry about inequalities between members of the group? ii) Local vs global justice -​ A further question concerns whether the norms of distributive equality (whatever they are) apply to all individuals, regardless of where they live. Or rather, do they only hold for members of communities within states and nations? Most theories of equality deal exclusively with distributive equality among people in a single society. -​ There does not, however, seem to be any rationale for that limitation. The universal morality of equal respect and the principle of equal distribution demand that all 19  persons consider one another as prima facie equally entitled to the goods, unless reasons for an unequal distribution can be advanced. -​ It may be that in the process of justification, reasons will emerge for privileging those who were particularly involved in the production of a good, but there is no prima facie reason to exclude from the outset other persons, such as those from other countries, from the process of distribution and justification. -​ That may seem most intuitively plausible in the case of natural resources (e.g. oil) that someone discovers by chance on or beneath the surface of his or her property. Why should such resources belong to the person who discovers them, or on whose property they are located? -​ Nevertheless, in the eyes of many if not most people, global justice, i.e., extending egalitarian distributive justice globally, demands too much from individuals and their states. -​ Alternatively, one might argue that there are other “special relations” between members of one society that do not exist between members of different societies. Nationalism is an example for such a (controversial) thesis that may provide a case for a kind of local equality. iii) Intergenerational justice -​ A further question concerns whether the norms of distributive equality (whatever they are) apply to all individuals, regardless of when they live. This raises the question of the relationship between generations. -​ Does the present generation have an egalitarian obligation towards future generations regarding equal living conditions? -​ One argument in favor of this conclusion might be that people should not end up unequally well off as a result of morally arbitrary factors. -​ However, the issue of justice between generations is notoriously complex. 2)​ EQUALITY WHEN? i) Starting-gate principles -​ Many distributive principles identify and require that a particular pattern of distribution be achieved or at least be pursued as the objective of distributive justice. But they also need to specify when the pattern is required, that is, the time frame. -​ One option is to require that all people should have the same goods (e.g. wealth) at some initial point, after which people are free to use their goods in whatever way they choose, with the consequence that future outcomes are bound to be unequal. -​ Principles specifying initial distributions after which the pattern need not be preserved are commonly called “starting-gate” principles. ii) Equality in time-frames -​ Since starting-gate principles may eventually lead to large inequalities, egalitarians have proposed different accounts. 20 -​ For example, the most common form of strict equality principle specifies that income (measured in terms of money) should be equal in each time-frame, though even this may lead to significant disparities in wealth if variations in savings are permitted. Hence, strict equality principles are commonly conjoined with some society-wide specification of just saving behavior. 3)​ WHY DOES EQUALITY MATTER? i) Intrinsic egalitarianism -​ Intrinsic egalitarianism is the thesis that equality is a good in itself. It is intrinsically bad if some people are worse off than others through no fault of their own. -​ Intrinsic egalitarians regard equality as desirable even when the equalization would be of no use to any of the affected parties, such as when equality can only be produced through depressing the level of well-being of everyone’s life. i) Intrinsic egalitarianism the levelling-down objection -​ This leads intrinsic egalitarianism to the famous levelling-down objection. Equality does not seem desirable if the only way to equalise is to bring everyone down without benefiting anyone. Doing away within equality in fact ought to produce better circumstances; it is otherwise unclear why equality should be desired. -​ Sometimes inequality can only be ended by depriving those who are better off of their resources, rendering them as poorly off as everyone else. This would have to be an acceptable approach according to the intrinsic conception. -​ But would it be morally good if, in a group consisting of both blind and sighted persons, those with sight were rendered blind because the blind could not be offered sight? That would be morally perverse. Doing away with inequality by bringing everyone down contains – so the objection goes – nothing good. i) Intrinsic egalitarianism and pluralistic egalitarianism -​ Such levelling-down objections would of course only be valid if there were indeed no better and equally egalitarian alternatives available, but there are nearly always such alternatives: e.g. those who can see should have to help the blind, financially or otherwise. -​ When there are no alternatives, in order to avoid such objections, intrinsic egalitarianism cannot be strict, but needs to be pluralistic. Pluralistic egalitarians do not have equality as their only goal; they also admit other values and principles, above all the principle of welfare, according to which it is better when people are doing better. ii) Instrumental egalitarianism -​ Instrumental egalitarianism is the thesis that equality is valuable only on instrumental grounds because it helps produce some desirable outcome. 21 -​ For example, helping the poor, promoting economic growth, or addressing social unrest and volatility. -​ The most immediate outcome of redistributing income and wealth from the rich to the poor is that the poor can get out of poverty. This is an instrumental justification of equality because the inequality itself matters less than helping people to get out of poverty. -​ A more equal distribution of resources can also serve other goals, such as promoting economic growth. Economic growth may be affected if there are too many people who are unable to invest in education or have lower health levels because of their poverty. This is a problem if we care about the well-being of the poor, but also if we care about economic growth. -​ Similarly, if inequality squeezes the middle class, it may also reduce its demand for goods and services and its capacity to invest in human capital and education. Finally, large wealth gaps can be associated with social conflict and instability because it makes it harder for society to come to a political consensus. -​ Yet, these instrumental arguments in favour of a more equal distribution of resources are more likely to be compatible with some level of inequality. As long as the poor get some support and inequality is not too large to damage the economic growth or social stability, some level of inequality may be acceptable. iii) Constitutive egalitarianism -​ Instrumental egalitarianism must be distinguished from a third justification for an equal distribution, namely constitutive egalitarianism. -​ The latter takes equality as having a constitutive value in the sense that its intrinsic value derives not from itself, but from being a part of a larger framework that has intrinsic value itself. This framework makes equality non-instrumental. -​ Thus, if we take justice as an intrinsic good, and part of what makes a social system a just one is that persons must have an equal claim to some goods, then equality can be taken to be a constitutive component of justice. 4)​ WHAT SHOULD BE EQUAL? i) Income -​ Income is the most straight-forward example of a resource metric to evaluate how well people are doing. In contemporary market economies, income is a very polyvalent good that gives people access to a large variety of external goods. -​ Using income as a metric simplifies many problems. It gives people the choice and avoids making controversial moral judgments about what should matter in people’s lives. It also facilitates the implementation and monitoring of distributive policies. -​ A challenge of using this approach is that people have different capabilities to convert resources into well-being. Depending on their personal abilities or handicap, a person may not be able to derive the same well-being or opportunities. 22 -​ Treating people fairly may require giving them a bit more than the others to make sure that they attain the same level of well-being or opportunities as others. ii) Opportunities and luck egalitarianism -​ The metric of opportunities is defended by so-called luck egalitarians. -​ It asks for people to be given equal opportunities, equal chances to succeed in life, and for compensating accidents in life. ii) Opportunities and responsibility -​ Luck egalitarians hold individuals responsible for their choices and actions, but not for circumstances beyond our control. The principle of responsibility provides a central normative vantage point for deciding on which grounds one might justify which inequality. -​ The positive formulation of the responsibility principle requires an assumption of personal responsibility and holds that inequalities which are the result of self-chosen options are just. Unequal portions of social goods are thus fair when they result from the decisions and intentional actions of those concerned. Persons are themselves responsible for certain inequalities that result from their voluntary decisions, and they deserve no compensation for such inequalities, aside from minimal provisions in cases of dire need. -​ In its negative formulation, the responsibility principle holds that inequalities which are not the result of self-chosen options are to be rejected as unjust; persons disadvantaged in this way deserve compensation. That which one can do nothing about, or for which one is not responsible, cannot constitute a relevant criterion. ii) Opportunities and formal equality of opportunities -​ A difficulty of this approach lies in identifying which inequalities stem from personal choice and which one arise from circumstances. -​ Formal equality of opportunity rules out formal discrimination on grounds such as a person’s race, ethnicity, age or gender. -​ Traits such as a persons’ gender or race are elements over which people have no control and, hence, a society in which people’s race or gender have fundamental effects on their lifetime economic prospects treats people unfairly. In such societies, whether people were born as the favoured gender or race, and hence were favoured economically, would simply be a matter of luck. ii) Opportunities and a more substantial form of equality of opportunities -​ The foregoing is relatively uncontroversial, but what makes these arguments so interesting is their claim that this line of reasoning actually leads to much stronger (and more contentious) requirements for social justice. -​ Even with formal equality of opportunity, there will remain many factors over which people have no control but which will affect their lifetime economic prospects, such 23  as whether a person’s family can afford to purchase good quality educational opportunities or health care. -​ A society therefore will have reasons to adopt a more substantial equality of opportunity principle, with equal opportunities for education, health care, etc. – the same reasons it had for adopting a merely formal equality of opportunity principle. ii) Opportunities and a radical form of equality of opportunities -​ Following this line of reasoning further seems to lead to more radical conclusions than those who agreed with formal equality of opportunity would have imagined. A society with a more substantial equality of opportunity principle in place will still not be providing equality of opportunity for all. -​ People are born into more or less nurturing families and social circumstances. People are born into families and neighborhoods which are more or less encoura- ging of education and the development of economically advantageous talents. -​ There are a whole range of social influences which have fundamental and unequal effects on children’s economic prospects and for which they are in no way responsible – the influences children are exposed to are a matter of their luck in the social lottery. -​ Moreover, the luck of the natural lottery is not just restricted to such characteristics as gender and race. Children are more or less fortunate in the distribution of natural talents as well. ii) Opportunities and the race metaphor -​ A race where the starting line is arbitrarily staggered, where people’s prospects for winning are not largely determined by factors for which they are responsible but rather largely by luck, is not considered a fair race. -​ Similarly, if society is structured so that people’s prospects for gaining more economic goods are not largely determined by factors for which they are responsible but rather largely by luck, then the society is open to the charge of being unfair. iii) Well-being -​ Welfarists propose instead that what matters is that everyone should have access to similar levels of well-being. But the welfarists have to be sophisticated about what should count towards well-being. -​ Historically, utility, the terms used by utilitarians instead of “welfare”, has been defined variously as pleasure, happiness, or preference-satisfaction, etc -​ Jeremy Bentham (1748–1832), the historical father of utilitarianism, argued that the experience of pleasure was the only thing with intrinsic value, and all other things had instrumental value insofar as they contribute to the experience of pleasure or the avoidance of pain. -​ His intellectual successor, John Stuart Mill (1806–1873), broadened this theory of intrinsic value to include happiness, or fulfillment. 24  iii) Well-being and preference-satisfaction maximization -​ Modern philosophers since Kenneth Arrow (1921–2017), though, tend to argue that intrinsic value consists in preference-satisfaction, i.e. in individuals’ having what they want. -​ So, for instance, the principle for distributing economic benefits for preference utilitarians is to distribute them so as to maximize preference-satisfaction. -​ The welfare function for such a principle has a relatively simple theoretical form requiring the distribution maximizing the arithmetic sum of all satisfied preferences (unsatisfied preferences being negative), weighted for the intensity of those preferences. -​ To accommodate uncertainty with respect to outcomes the function is modified so that expected utility, rather than utility, is maximized. iii) Well-being and the first critique -​ The first critique, which was famously articulated by John Rawls (1921–2002), is that utilitarianism fails to take seriously the distinctness of persons. -​ Maximization of preference-satisfaction is often taken as prudent in the case of individuals – people may take on greater burdens, suffering or sacrifice at certain periods of their lives so that their lives are overall better. -​ The complaint against utilitarianism is that it takes this principle, commonly described as prudent for individuals, and uses it on an entity, society, unlike individuals in important ways. -​ While it may be acceptable for a person to choose to suffer at some period in her life (be it a day, or a number of years) so that her overall life is better, it is often argued against utilitarianism that it is immoral to make some people suffer so that there is a net gain for other people. -​ In the individual case, there is a single entity experiencing both the sacrifice and the gain. Also, the individuals, who suffer or make the sacrifices, choose to do so in order to gain some benefit they deem worth their sacrifice. -​ In the case of society as a whole, there is no single experiential entity – some people suffer or are sacrificed so that others may gain. -​ Furthermore, under utilitarianism, unlike the individual prudence case, there is no requirement for people to consent to the suffering or sacrifice, nor is there necessarily a unified belief in the society that the outcome is worth the cost. iii) Well-being and the second critique -​ A related criticism of utilitarianism involves the way it treats individual preferences about other peoples’ welfare or holdings. -​ For instance, some people may have a preference that the members of some minority racial group have less material benefits. -​ Under utilitarian theories, in their classical form, this preference or interest counts like any other in determining the best distribution. 25 -​ Hence, if racial preferences are widespread and are not outweighed by the minority’s contrary preferences (perhaps because the minority is relatively few in number compared to the majority), utilitarianism will recommend an inegalitarian distribution based on race if there is not some other utility-maximizing alternative on offer. 26 Lecture 2 FIFTH QUESTION: BATTLE OF DISTRIBUTION 3 POSSIBLE ANSWERS 5) How should goods be distributed? i) Sufficientarianism: it's a theory it says we have to alleviate the suffering. -​ Among the least demanding conception of justice, sufficientarianism merely holds a humanitarian concern in helping those worse off and alleviating suffering. -​ Positive thesis: we have reasons to secure enough that are weighty and non-instrumental. Negative thesis: once individuals have secured enough we no longer have reasons, of justice at least, to be concerned with the distribution of benefits and burdens from their perspective. -​ In his book Equality as a Moral Ideal (1987), which is considered the seminal work of modern sufficientarianism, Harry Frankfurt claims that egalitarian thought contributes to the moral shallowness of our time and diverts our attention away from considerations of greater importance than equality. -​ More positively, he argues that if everyone had enough it would be of no moral consequence whether some had more than others’. i) Sufficientarianism and the Indifference Objection -​ Sufficientarianism is not understood as egalitarian because the goal is not to reduce relative differences between people, but to bring the absolute amount of resources, opportunities, or well-being of the worst off above the sufficient level. -​ One issue with sufficiency, which is called the Indifference Objection, is that it is compatible with very large inequalities, including those that are not deserved at all and are the result of the natural lottery of talents, place of birth, or social background. -​ As long as we bring the worst off above the sufficiency level, large inequalities can remain. ii) Strict egalitarianism: second answer. (it is the idea that everyone should be furnished the same material level of goods and services. It’s usually unattainable, can’t be put into practice. U associate to communism’s or socialism, however this is not the case, cause either comunis or socially are for absolute equal economy. ) 27  -​ Strict egalitarianism, namely the idea that everyone should be furnished the same material level of goods and services, is generally rejected as untenable. Hence, with rare possible exceptions, no prominent author or movement has demanded strict equality. -​ Since egalitarianism has come to be widely associated with the demand for economic equality, and this in turn with communistic or socialistic ideas, it is important to stress that neither communism nor socialism calls for absolute economic equality. ii) Strict egalitarianism and communism The orthodox Marxist view of economic equality was expounded in the Critique of the Gotha Program (1875). Marx here rejects the idea of legal equality, on three grounds. -​ i. Equality draws on a limited number of morally relevant perspectives and neglects others, thus having unequal effects. The economic structure is the most fundamental basis for the historical development of society and is thus the point of reference for explaining its features. -​ ii. Theories of justice have concentrated excessively on distribution instead of the basic questions of production. -​ iii. A future communist society needs no law and no justice, since social conflicts will have vanished. ii) Strict egalitarianism and objections Among the objections against strict egalitarianism, we find: -​ 1 The levelling-down objection (see Lecture 1); -​ 2 Equality distorts incentives promoting achievement in the economic field, and the administrative costs of redistribution produce wasteful inefficiencies. Equality and efficiency need to be balanced. -​ 3 A strict and mechanical equal distribution between all individuals does not sufficiently take into account the differences among individuals and their situations. Individuals desire different things, so why should everyone get the same things. In essence, since individuals desire different things, why should everyone receive the same goods? -​ 4 There is a danger of (strict) equality leading to uniformity, rather than to a respect for pluralism and democracy. In the contemporary debate, this complaint has been mainly articulated in feminist and multiculturalist theory. iii) The maximin principle and the veil of ignorance. A different proposal is offered by John Rawls in his book A Theory of Justice (1971). It’s a masterpiece in the philosophy of justice Rawls develops his principles of justice through the use of an artificial device, the original position, in which, everyone decides principles of justice from behind a veil of ignorance. This veil is one that es- sentially blinds people to all facts about themselves so they cannot tailor principles to their own advantage: 28  [N]o one knows his place in society, his class position or social status, nor does anyone know his fortune in the distribution of natural assets and abilities, his intelligence, strength, and the like. I shall even assume that the parties do not know their conceptions of the good or their spe- cial psychological propensities. The principles of justice are chosen behind a veil of ignorance. Ignorance of these details about oneself will lead to principles that are fair to all. iii) The maximin principle and Rawls’ principles of justice Rawls proposes the following two principles of justice: -​ 1. Each person has an equal claim to a fully adequate scheme of equal basic rights and liberties, which scheme is compatible with the same scheme for all; and in this scheme the equal political liberties, and only those liberties, are to be guaranteed their fair value. -​ 2. Social and economic inequalities are to satisfy two conditions: (a) They are to be attached to positions and offices open to all under conditions of fair equality of opportunity; and (b), they are to be to the greatest benefit of the least advantaged members of society. Where the rules may conflict in practice, Rawls says that Principle (1) has lexical priority over Principle (2), and Principle (2a) has lexical priority over (2b). While it is possible to think of Principle (1) as governing the distribution of liberties, it is not commonly considered a principle of distributive justice given that it is not governing the distribution of economic goods per se. We have already discussed the principle of equality of opportunity, so we now focus on (2b), which is known as the Difference Principle. iii) about the effect of incentives and the benefits that may flow to all from the productive labours of the most talented members of society, the maximin principle is consistent with a considerable degree of inequality. -​ These inequalities can benefit everyone and help maximise the share of the worst off compared to strict equality. This provides a powerful argument for tolerating some inequality, because everyone benefits from it, even the worst off. 29  AI AND WORK Global inequalities have increased worldwide in the past decades. The causes of this process are complex. One important driver of rising inequality is technological change. How can AI contribute to this phenomenon? In several ways. One of them is the impact of AI in the world of work. -​ AI for recruitment -​ Building, maintaining, and testing AI systems -​ AI and the labour market -​ AI at the workplace Strategeion: a case study -​ A small, enthusiastic group of Army veterans, trained in programming and com- puter engineering, co-founded the non-profit company Strategeion after having been honorably discharged. Driven by a strong sense of civil virtue, Strategeion was an online platform offering a variety of services of public utility, which were primarily dedicated to veterans - but also to other people, according to the motto “leave no one behind”. -​ Whereas other innovative tech firms mostly employ young, recent graduates from prestigious universities, at the beginning Strategeion was staffed largely by ex- military personnel. The high employees satisfaction and retention originated from this policy resulted in a surge of public visibility and job inquires. The number of applications had far outpaced the number of positions available, more and mo- re of them were coming from the kind of traditional candidates that might have typically applied for jobs at larger, for-profit tech companies. Efficiency -​ The human resources (HR) team became so overwhelmed that a group of Strate- geion’s developers decided to implement a system, called PARiS, to automatically pre-sort resumes according to a candidate’s fit within the company. -​ PARiS used natural language processing and machine learning to look for markers in resumes that distinguished the best candidates. In order to train the system, HR provided the engineering team with dozens of resumes from current and previous employees who were deemed either exemplary or especially poor in terms of professional attributes and fit. PARiS would rate incoming resumes according to their match with the ideal types and cast aside those that were below a set threshold. -​ Over time, growing trust in the system meant that the HR representatives felt less and less need to double-check PARiS’ work. 30  Question 1: Paris promised to make the hiring process more efficient. But are there other values that might be desirable in hiring? Diversity? Equity? Creativity? What, if anything, do companies risk losing when hiring procedures are so singularly focused on maximizing efficiency? Strategeion: a case study -​ Hara, a promising and hard-working computer science student from Athens, received an automated rejection email from Strategeion within hours of applying for a job through its website. She had strong academic qualifications and she had carefully crafted her resume to reflect her civic commitments and experience working with non-profit organizations that advocated for wheelchair users such as herself. Disappointed at her rejection, Hara wrote to the company asking for feedback on her application. -​ Hara’s request made its way to the HR department, and the representative who received it was also puzzled by her rejection. After having thoroughly reviewed. Hara’s application, he judged her to be on par with Strategeion’s very best employees in terms of both interests and credentials. -​ One potential concern was that PARiS may have used Hara’s disability status as a reason to deny her application. However, the system’s engineers had explicitly designed the algorithm so that it would not discriminate against protected categories. Responsibility -​ Furthermore, Strategeion’s policy of hiring ex-military personnel – many of whom were wheelchair users – meant that the system’s training data was not biased against those with physical disabilities. But if it wasn’t her disability, then what was it that PARiS had found in Hara’s resume that had caused it to categorize her as a bad fit? -​ The unlikely answer was: sports. There was a strong positive correlation between participation in athletics and military service. Given the overrepresentation of veterans among Strategeion’s employees and their propensity to excel at the company, PARiS had learned to connect a history of playing sports with good fit. While it was true that many of Strategion’s ex-military employees no longer participated in sports, their resumes typically reflected a history of having done so. Hara, having used a wheelchair her entire life, had no history of athletic activities. Question 2: Biased data sets pose a problem for ensuring fairness in AI systems. What could Strategeion’s engineers have done to counteract the skewed employee data? To what extent are such efforts the responsibility of individual engineers or engineering teams? -​ The HR representative in charge of Hara’s case reached out to her with the team’s findings. The HR agent apologized on behalf of the company and invited Hara for an interview. Hara ultimately decided to reject Strategeion’s offer of an interview and, 31  instead, she filed an official complaint with the company, in which she also specifies that: -​ Fairness. Hara has been treated unfairly, because PARiS had ultimately discriminated against her application on the basis of an irrelevant characteristic (i.e., her physical capabilities). Moreover, given the history of marginalization and the lack of accommodations traditionally made for persons with disabilities, the fairest thing for Strategeion to have done would have been to engineer PARiS to positively discriminate in favor of those with physical disabilities. -​ Dehumanizing system. Hara was disconcerted by the idea of a non-human agent deciding whether she’d even have the opportunity to make her case in a job interview. For a decision that important, she argued that there ought to be a human in the loop. The process of being converted to an input and assessed in this manner can feel dehumanizing. Hara even suggested that the sense of dehumanization may extend to the HR workers who had a central aspect of their job replaced by a machine. -​ Consent and contextual integrity. Hara was dismayed that automated decision-making tools had been used to evaluate her resume without her explicit consent. And she wasn’t alone. Upon learning about PARiS, many of Strategeion’s current and former employees were unhappy that their resumes might have been used to train the underlying datasets without their knowledge or permission. DISCRIMINATION Strategeion first addressed the legal allegations in Hara’s complaint. If PARiS had been directed to discriminate against applicants on the basis of their disability status, Strategeion would clearly have violated US law. But that was not the case. PARiS was not intentionally discriminating against resumes based on protected attributes; rather, the system had inferred such attributes from other, seemingly innocuous data. Thus, Strategeion’s lawyers believed they could prove the company was legally in the clear. Question 3: The type of discrimination practiced by PARiS might not seem as blatantly demeaning as a blanket hiring policy against those with physical disabilities, but it is any different from a moral standpoint? How might this kind of insidious discrimination, which is, by definition, difficult to spot, be avoided? HOMOGENEOUS vs HETEROGENEOUS WORKFORCE One option to address PARiS’ bias problem called for rethinking the value of a homogenous workforce. Recent reports in management studies have shown that more diverse project teams are able to evaluate products and services from a wider range of perspectives, typically resulting in all-around better outputs, as well as more productive workplaces. Upon 32 reading some of this literature, even Strategeion’s co-founders tentatively agreed that it might be worth considering a change in hiring priorities. Question 4: Social science increasingly shows that there are advantages to a heterogenous workforce, but there are also advantages to homogeneity. A diverse workforce helps protect organizations against “group think,” for example, but groups that share certain experiences and backgrounds may find it easier to communicate with and understand one another, thereby reducing collective action problems. If you were a manager in charge of hiring at Strategeion, for which position would you advocate? Would you try to maintain the corporate culture by hiring people who resemble current employees, or would you argue that PARiS should be realigned to optimize for a broader range of types? introduction: HISTORY OF HIRING TECHNOLOGY Hiring technology has evolved rapidly alongside the internet. As early as the 1990s: ▶ online job boards (Monster.com) offered digital job listings (at rates well below those of newspaper classified ads); ▶ search engines for these online job postings emerged soon after; and ▶ pay-per-click advertising helped recruiters compete for attention. -​ Next came new ways to apply for jobs over the internet, and it became easier to apply for multiple jobs. -​ Meanwhile, recruiters began using digital technology to proactively seek out desirable applicants. By scouring new, public sources of information (like professional profiles like LinkedIn), recruiters were able to broaden their focus from active candidates to passive ones. -​ As the quantity of potential job candidates exploded, some employers began turning to new screening tools to keep up. While employers had long relied on tests and assessments to screen jobseekers, the development of new techniques to collect and analyze data prompted the introduction of more advanced assessments. -​ In response to the growing push for diversity and inclusion in the workplace, some technology vendors have more recently introduced tools to facilitate diversity recruiting and reduce various biases endemic to the hiring process. -​ Today, hiring technology vendors increasingly build predictive features into tools that are used throughout the hiring process. They rely on machine learning techniques, where computers detect patterns in existing data to build models that forecast future outcomes in the form of different kinds of scores and rankings. 33 Why employers adopt predictive tools -​ Most employers want to reduce time to hire (i.e., the amount of time it takes to fill an open position). ▶ the longer it takes to find a suitable candidate, the more time and resources are diverted from other priorities; ▶ employers also fear losing candidates to their competitors; ▶ some companies have seasonal staffing needs that make it critical to hire new employees within a particular time frame. -​ Employers also want to reduce cost per hire, or the marginal cost of adding a new worker, which is roughly $4,000 in the U.S. -​ Employers also try to maximize quality of hire, which is judged based on metrics related to performance evaluations, the quantity or quality of worker output, or whether the hire was eventually promoted or disciplined. -​ Many employers also look to maximize the tenure of their workers, presuming that successful hires will stay longer than less successful ones. Turnover is costly, requiring an employer to hire and train new workers. -​ Finally, some employers have goals for workplace diversity, based on gender, race, age, religion, disability, or veteran or socioeconomic status. Discrimination and bias DIRECT (OR INTERPERSONAL) DISCRIMINATION -​ Hiring tool vendors often tout technology’s potential to remove bias from the hiring process. They argue that their tools make hiring more consistent and efficient, make fairer and more holistic hiring decisions, or naturally reduce discrimination by obscuring applicants’ sensitive characteristics. -​ Vendors are usually referring to direct or interpersonal discrimination, which happens when protected attributes of individuals explicitly result in non-favorable outcomes toward them. Typically, there are some traits identified by law on which it is illegal to discriminate against, and it is usually these traits that are considered to be “protected” or “sensitive” attributes in computer science literature. -​ A list of some of these protected attributes is specified in the Fair Housing and Equal Credit Opportunity Acts: race or color, national origin, religion, sex, familiar status, disability, marital status, recipient of public assistance, age. -​ Direct or interpersonal discrimination is only one source of discrimination. INDIRECT (OR SYSTEMIC) DISCRIMINATION -​ Institutional discrimination arises at the institutional level when policies and workplace cultures serve to benefit certain workers and disadvantage others. For example, a company that tends to hire from a privileged and homogeneous 34  community and then uses “culture fit” as a factor in hiring decisions could end up methodically rejecting otherwise qualified candidates who come from more diverse backgrounds. -​ Hiring practices can also perpetuate structural discrimination: patterns of disadvantage stemming from contemporary and historical legacies such as racism, unequal economic opportunity, and segregation. For example, many white collar employers place a high value on elite university attendance, but despite changing admissions policies, such a credential is still disproportionately attained by privileged individuals, and often out of reach for those who lack access to quality primary and secondary education. -​ Discrimination can also be internalized by jobseekers themselves, influencing their own behaviors, such as whether or not to apply for a given job. A TERMINOLOGICAL OBSERVATION -​ How can predictive tools perpetuate discrimination? Through bias, which can exist and emerge in predictive tools in several distinct ways. -​ Similar to discrimination, bias is also a source of unfairness. Discrimination can be considered as a source for unfairness that is due to human prejudice and stereotyping based on the sensitive attributes, which may happen intentionally or unintentionally, while bias can be considered as a source for unfairness that is due to the data collection, sampling, and measurement. -​ But this distinction is not very strict (e.g., bias can also be seen as a source of unfairness that is due to human prejudice and stereotyping). THE DATA, ALGORITHM, AND USER INTERACTION FEEDBACK LOOP 1) Most AI systems and algorithms are data driven and require data upon which to be trained. In the cases where the underlying training data contains biases, the algorithms trained on them will learn these biases and reflect them into their predictions. 2) As a result, existing biases in data can affect the algorithms using the data, producing biased outcomes. Algorithms can even amplify and perpetuate existing biases in the data. In addition, algorithms themselves can display biased behavior due to certain design choices, even if the data itself is not biased. 3) The outcomes of these biased algorithms can then be fed into real-world systems and affect users’ decisions, which will result in more biased data for training future algorithms. -​ Amazon’s experimental hiring tool used artificial intelligence to give job candidates scores ranging from one to five stars – much like shoppers rate products on Amazon. -​ But by 2015, the company realized its new system was not rating candidates for software developer jobs and other technical posts in a gender-neutral way. 35  -​ That is because Amazon’s computer models were trained to vet applicants by observing patterns in resumes submitted to the company over a 10-year period. Most came from men, a reflection of male dominance across the tech industry. -​ In effect, Amazon’s system taught itself that male candidates were preferable. It penalized resumes that included the word “women’s,” as in “women’s chess club captain.” And it downgraded graduates of two all-women’s colleges. THREE KIND OF BIAS Bias can exist in many shapes and forms, and they can be classified according to the data, algorithm, and user interaction loop: -​ Data-to-algorithm bias: biases in data, which, when used by algorithms, might result in biased algorithmic outcomes. -​ Algorithm-to-user: biases that are as a result of algorithmic outcomes and affect user behavior as a consequence. -​ User-to-data: many data sources used for training ML models are user-generated. Any inherent biases in users might be reflected in the data they generate. 1)​ Data-to-algorithm bias i) Omitted-variable bias Definition (omitted-variable bias) Omitted-variable bias occurs when one or more important variables are left out of the model. Example: Consider a model designed to predict the annual percentage rate at which customers will stop subscribing to a service. Suppose that the majority of users are canceling their subscription without any warning from the designed model. Imagine that the reason for canceling the subscriptions is appearance of a new strong competitor in the market which offers the same solution, but for half the price. The appearance of the competitor was something that the model was not ready for; therefore, it is considered to be an omitted variable. ii) Representation bias Definition (representation bias): Representation bias arises from how we sample from a population during a data collection process. Example: ImageNet is an image database instrumental in advancing computer vision and deep learning research. The fraction of US and Great Britain are the top represented locations. This results in demonstrable bias towards Western cultures. 36  iii) Aggregation bias