Using Data Responsibly

Better Data for Doing Good: Responsible Use of Big Data and Artificial Intelligence

Introduction

The data universe is ever expanding, as chapter 2 illustrates. In fact, it is estimated to double in size every two years with some 2.5 quintillion bytes of information being generated daily. Because we increasingly use digital devices to communicate, buy and sell goods, transfer money, search for information on the internet, and share our lives on social networks, we leave digital trails or "digital exhaust". A growing amount of digital data is thus being generated as a by-product of our daily lives, but also through the increasing digitization of content and the spread of the Internet of Things. This growing volume of data is driving the development of big data analytics and artificial intelligence (AI), the subjects of this chapter. The chapter describes opportunities for harnessing the value of big data and AI for social good, and how new families of AI algorithms now make it possible to obtain actionable insights automatically and at scale. Beyond internet business or commercial applications, multiple examples already exist of how big data and AI can help us achieve our shared development objectives, such as the 2030 Agenda for Sustainable Development and the Sustainable Development Goals (SDGs). But ethical frameworks need to be developed in line with increased uptake of these new technologies – any discussion of ethics is not limited to the privacy of the data, but also relates to the impact and consequences of using data and algorithms – or failing to use them.

Source: World Bank, https://openknowledge.worldbank.org/handle/10986/30437
This work is licensed under a Creative Commons Attribution 3.0 IGO License.

The Big Data Revolution

As chapter 1 notes, the concept of big data typically describes data sets so large, or so complex, that traditional data processing techniques often prove inadequate. The term "big data" thus captures not only the large volumes of data now available, but also the accompanying processes and technologies for collecting, storing, and analyzing it. In other words, "big data" is not just about data – "no matter how big or different it is considered to be" – it is primarily about "the analytics, the tools and methods that are used to yield insights," including the frameworks, standards, and stakeholders involved in the field and ultimately the knowledge generated.

Although businesses increasingly are mining the digital trails we leave behind to predict consumer behavior, track emerging trends in the market, and monitor operations in real time to improve sales and profit margins, big data analytics also holds enormous potential to help understand and address pressing socioeconomic and environmental issues.

Big data can help inform policy and interventions that set us on a more sustainable development path and improve responses to humanitarian emergencies.

Innovation labs across academia, government, the international development community, civil society, and the private sector have been using big data and AI to develop a wide range of applications, from mapping discrimination against refugees in Europe to facilitating the rescue of migrants at sea based on shipping data, detecting fires in the Indonesian rainforest, predicting food insecurity due to changing food prices via Twitter, or fighting the effects of climate change. Box 3.1 describes how big data is also being used to predict and respond to disease outbreaks.

The Evolution of Artificial Intelligence

Historically, the term "artificial intelligence" has been applied where computer systems imitate thinking or behavior that people associate with human intelligence, such as learning, problem solving, and decision-making. Modern AI comprises a rich set of subdisciplines and methods that leverage technologies such as visual, speech, and text recognition, as well as robotics. Machine learning is one such subdiscipline. Whereas hand-coded software programs typically contain specific instructions on how to complete a task, machine learning allows a computer system to recognize patterns and make predictions. Deep learning, a subset of machine learning, goes one step further ­– with deep artificial neural networks, based on complex algorithms, computers can learn from large volumes of data while reaching new levels of accuracy. 

In sum, AI is enabling computer systems to collect, analyze, and process large amounts of data in real time to recognize patterns, make decisions, and, more significantly, to learn from said data and from their own experiences.

Meanwhile, recent advances in sensors and imaging technologies and data storage, processing, and transfer technologies, as well as complex and self-improving algorithms, to name but a few, are the range of expanding AI applications available today. AI is already incorporated in several online products, including Google search, Google Translate, and Facebook's automatic photo-tagging and translation applications. Financial companies rely on AI to produce the financial modeling that underpins their insurance, banking, and asset management products. Moreover, leading research hospitals have started using AI tools to help medical professionals diagnose and choose the best course of treatment for their patients.

Although the current application of AI is mostly limited to internet business, digital marketing, gaming, and selfdriving cars, a wealth of opportunities exist for AI methods to perform different tasks that can accelerate achievement of the SDGs and inform humanitarian practice. Box 3.2 describes how AI can help transform traditional sectors, such as transport.

Using Big Data and AI as a Force for Social Good

AI and big data are generating new tools and applications creating actionable insights, real-time awareness, and predictive analysis on numerous topics for sustainable development and humanitarian action. More and more compelling examples illustrate the value of this technology to improve early warning systems and inform policy and programmatic response. These individual use cases represent a small but significant innovation in learning about the world around us. Taken together, they provide new ways to detect and respond to world events, influence policy debates, and drive development, in a way that is both safe and fair (figure 3.1, table 3.1). 

The following sections examine the benefits and applications of big data and AI ­– including for (a) speech and audio processing, (b) image recognition and geospatial analysis, and (c) text analysis. They also describe how AI is being leveraged to support the SDGs and address the emerging challenges and risks that accompany the uptake of these technologies.

Figure 3.1 The Sustainable Development Goals

Table 3.1 Examples of artificial intelligence applications for the Sustainable Development Goals

Speech and audio processing

Arguably, one major achievement of big data and AI has been to facilitate real-time translation of a growing number of the world's languages. Although language translation is not an SDG per se, greater language and cultural understanding could help increase the efficiency and effectiveness of development efforts across all SDGs ­– for example, by helping to map public opinion (see box 3.3). Google and Microsoft systems, for example, are now able to translate over a hundred languages. Also, new systems have been developed that perform real-time translations ­– such as a Skype system that can translate voice calls into 10 different languages in real time.

Early models of machine translation used statistical methods that translated words based on a short sequence, that is, within the context of several words before and after the target word, which did not always work for long and complex sentences. New neural network architectures, such as long short-term memory, have drastically improved efficiency. Such systems can now learn from millions of examples and are able to translate whole sentences at a time, rather than word by word.

Computer vision, image analysis, and geospatial data

Accurate population information is critical for authorities to plan and deliver quality public services and coordinate crisis-relief efforts. However, collecting related data traditionally is a long-standing challenge for development practitioners and policy makers. For example, gathering national household survey data on poverty is typically time-consuming and expensive, requiring elaborate data collection and analysis techniques. This exercise is particularly challenging in fragile states, where limited capacity and security concerns typically hinder data collection and processing. In this setting, for example, satellite imagery has been used to gain an overview of population density and assess poverty and access to energy ­– covered by SDG 1 and SDG 7 (see boxes 3.4 and 3.5).

In the health sector ­– covered by SDG 3 ­– current advances in medical imaging and computer analysis of tumors can complement and refine radiologists' analysis. Mobile phone call records have also been combined with satellite data to build dynamic population maps and estimate cross-border flows of migrants to enable development actors to track the spread of disease. This technique was leveraged in southern Africa to map the movements of cross-border communities to better understand malaria infections patterns.

In the environmental field ­– SDGs 12, 13, 14, and 15 ­– AI-assisted analysis of satellite imagery can be used to monitor damage to coastal areas due to floods or typhoons, or drought-affected areas, or the retreat of wetlands or encroaching land use in deltas or river basins. Combined with meteorological models and large data sets on changes in ocean temperature and currents, such mapping can help improve forecasting and early warning systems of future major weather events. Moreover, GPS data has been used to analyze traffic patterns to reduce pollution (see box 3.6). Another AI application getting considerable attention is automated or self-driving cars ­– a potential solution for optimizing transportation in ways that can minimize car accidents. Debate is ongoing about what a fully automated car really is, but considerable progress has been made toward solving problems of visual recognition, object identification, and reaction processing, which are critical to this endeavor.

Building on humble beginnings and minor innovations (including cruise control, assisted steering, lane assist, automatic braking, and "Traffic Jam Assist"), the race toward a fully automated car is now underway (box 3.7).

Text mining and text analysis

From Design to Responsible Use: Ethical Challenges with Using Big Data and AI

Although we are only scratching the surface of what is possible in the new age of big data and AI, and how they can be leveraged for social good, we also need to grapple with both the unintended risks and malicious use of the same technology. These benefits and looming risks were aptly articulated by the UN Secretary-General at the 2017 "AI for Good Global Summit":

A Way Forward: Harnessing Big Data and AI to "Leave No One Behind"

This chapter has detailed a handful of examples of the many innovative applications of big data and AI being used to inform sustainable development and humanitarian work globally (see table 3.1 in particular), illustrating the value of this technology for development actors.

The pervasive nature of big data and the rapidly evolving capabilities of AI hold tremendous promise for social impact and can drive transformation across many domains, ranging from health, to food security, to jobs, and action on climate. Scope therefore exists to expand use of this technology beyond current applications, leveraging big data and AI in new ways that help us achieve the 2030 Agenda. National and international development actors should prioritize operational integration of these digital innovations into policy and practice. Doing so will allow them to craft more agile and responsive programming, to support anticipatory approaches to managing risk, and to find new ways to mea sure social impact. However, mainstream, scaled adoption by policy makers and communities themselves still faces systemic barriers and pervasive inertia.

Given their broad applicability, big data and AI necessitate new forms of interinstitutional relationships to leverage data and computational resources, human talent, and decision-making capacity. The capabilities of a diverse set of stakeholders can enable the integration of data innovation into ongoing policy processes rather than one-time policy decisions.

Moreover, as adoption of big data and AI increases and the technology evolves, so do the potential risks and issues that need to be resolved. Many question the suitable application of this technology, including malicious use, and highlight the risk of unintended consequences in this rapidly evolving field, where policy makers may struggle to keep pace. Although both the supply of and demand for data are expanding at "warp speed," the data ecosystem, as we know it, is still embryonic – with many advanced potential applications still more theory than practice. As new capabilities and data sources are applied for good – whether to create smarter public services, better early warning systems, or more effective responses to crises – development actors must pause to consider the potential for harm that may arise, for example, from inadequate privacy protection.

To date, no standards exist for the anonymization and sharing of insights from big data in priority industries such as financial services, e-commerce, and mobile telecommunications – although the latter has done work to develop such standards. At the same time, as noted, nonuse of these new capabilities and data sources represents at least as great a risk of harm to the public as that potentially arising from inadequate privacy protections. New frameworks are needed that go beyond privacy and ensure accountability and responsible use and reuse of data for the public good. Principles such as responsibility, accuracy, auditability, and fairness should be core concepts that guide the development of algorithms and AI. The "society-in-the-loop" algorithm concept, for example, proposes to embed the "general will" into an algorithmic social contract in which citizens oversee algorithmic decision-making that affects them.

Developing countries may have the most to gain from the use of new data sources and tools. However, without thoughtful application and critical complements they may also stand to lose the most. To reap the societal benefits of AI – including expected improvements to productivity and innovation – countries must have access to the data, tools, and human expertise necessary to support their application, as well as viable plans to address the likely displacement of workers. The availability of data is to a large degree a by-product of digitization, an area in which developing countries lag far behind. There can be no mass digitization without universal and affordable access to broadband. According to International Telecommunication Union (ITU) statistics, some 3.8 billion people, or just over half the world's population, were still lacking access to the internet in 2017.

The way forward must be inclusive. For the big data and AI revolution to benefit the most vulnerable people, current AI research roadmaps must increase attention to methodologies that can work in data-scarce environments, that can be adapted quickly and with few examples – as in crisis scenarios – and that can work with incomplete or missing data (such as "one-shot learning"). Need is also urgent for bridging gender inequalities in big data. More effort must be made to train younger generations, women and men, to ensure gender equality and the inclusiveness of ethnic groups in shaping AI.

As the field of data science accelerates, countries must create robust big data and AI strategies to prevent growing inequalities in access and use of these technologies. In digital advertising, for example, where many of these capabilities were incubated, big data and AI continue to demonstrate their ability to concentrate wealth – and data – in the hands of the few and widen inequalities.

Just as misuse of AI may lead to harm, nonuse of AI may allow preventable harms to occur. The challenge is that misuse of these new tools is already rife online and real harm is being done, while the opportunity cost of failure to use them responsibly is mounting. Clearly, although achievement of the 2030 Agenda and the modernization of humanitarian practices necessitates responsible use of these new tools, it urgently requires a new, rights-centric effort by all stakeholders to ensure innovations meet community needs and no one is left behind. Undoubtedly, assessing the ethical impact of AI in addition to privacy protection measures can mitigate harm, maximize benefit, and lead to use of the new technologies as a force for good.

People and Data

Introduction

How can the data revolution expand economic development opportunities for more people? Can the increasing collection, analysis, and use of data – often from individuals1 through digital transactions or digital records of offline activities – broadly benefit those individuals and people? And what risks might arise, such as to individual privacy, and how might they be managed?

Conversations about data have become very popular: interest over time in "big data," as indicated by Google searches, for instance, has grown one-hundred-fold since 2010.2 More data is being generated – by people and machines – and captured, processed, and transferred than ever before. Much of this is because of the increasing use of digital technologies by people and organizations globally; indeed, even most analog processes have digital components (such as a visit to a doctor's office leading to a digital drug prescription).

But while the data revolution can benefit people, this chapter proposes that the structure of data markets might be raising risks and costs to individuals. People bear many of the costs and risks of participating in data markets and, indeed, might not even be aware they are participating. The poorest also face entry barriers, and it is possible that they might not benefit from their participation even when it is possible. 

The benefits from data include – at the most general level – the ability of data users to make better decisions using the information processed and to enjoy more convenience when interacting with organizations (for instance, easier interchange of data between platforms or service providers). The increasing use of such data by organizations, such as businesses and governments, implies the potential for faster and better decisions by these entities. This can help them improve service delivery, reduce costs and prices, or support process or product innovation, all of which would benefit the people that those organizations serve.

For example, better techniques for tracking how and where people drive their cars can inform traffic planning and management. Data from people's online activities inform advertising decisions that fund the operation of many widely used internet services that are "free" at the point of use. And as digital tools proliferate, individuals are increasingly able to benefit directly from access to more and new types of data and the information derived from it. People can take steps to increase their physical activity and improve their health by using digital pedometers, which have now become available on many smartphones apart from watch-like activity trackers. They can analyze market trends and make more informed choices about the products or services they buy, for instance, when buying books or purchasing air tickets. And depending on the organizations that use that data, people may benefit indirectly by being better 52 Information and Communications for Development 2018 able to navigate the organization's products and services and through an expanded set of choices or opportunities, based on analysis of the preferences exhibited by collating data on consumer and web traffic choices. 

The possible costs to the user of data collection include the loss of privacy, of agency, and of control. Such costs can undermine people's trust in the organizations that collect, control, and use data. Indeed, at the time of writing, various controversies had broken out over data leaks compromising the privacy of personal data and the biases involved in the use of data to profile individuals; these have underscored the risks emerging in the new data-rich economy. These costs are not always apparent or are distributed in biased ways among participants in data markets. This is because of how those markets have been evolving, with some organizations gaining significant power in defining how such data is collected, used, and shared. Other risks are emerging in this era of data: because of barriers that prevent people – especially the poor – from participating effectively in data markets and analog limitations to the benefits of the data revolution.

The chapter considers several aspects of personal data markets, which run on the personally identifiable data that people generate (figure 4.1 reviews the types of personal data). It looks at how data markets have evolved, highlights the various players in the data market, and then discusses the benefits and costs for participants in data marketplaces through digital networks and how negative impacts might be reduced. The chapter concludes with a discussion of public policies that could rebalance the costs and benefits to ensure fairer distribution among participants and understand how data marketplaces can focus more on people. These choices could determine whether data will help people – especially the poor – find economic opportunity. Few best practices exist as models, and hence the chapter will leave the reader not with specific policy prescriptions, but with a better sense of the dynamics at play.

The Data Market

Technological change and evolving business models

Personal data is generated through an individual's actions (such as making a payment using a credit card), through business processes that digitize analog data (such as medical histories), or through consequent machine response (such as call data records). Such data is now increasingly coming from use of the internet, wireless sensors, and the billions of mobile phones around the world. As the world gets more connected, more people are leaving a digital trail, wherever they go and whatever they do.

 This data, which has become more voluminous and granular over time, piqued the interest of various organizations that saw the financial value embedded in it. By the early 1990s, personal data such as telephone numbers and email addresses was widely used for marketing. Companies crunched data to predict how likely people would be to buy a product, and began using that knowledge to come up with targeted marketing messages. As more digital data was collected, organizations began to use increasingly powerful computing tools to manipulate and apply that data. Marketing companies built richer consumer profiles to predict future purchases and manufacturing and services companies to design and model new products.

Companies are now using such data to develop services powered by artificial intelligence (AI), and the bigger the data set, the better the AI. These and other innovations have greatly increased the value of data and its potential for being monetized, or bought and sold as a product in its own right. Data continues to gain value as its potential uses increase. Organizations – including businesses, governments, and others – can derive value from data by applying the insights arising from data's analysis to internal cost and revenue optimization, marketing and advertising, intelligence and surveillance, and automation.

The application of personal data for online advertising has skyrocketed, with the internet now surpassing television as the leading advertising channel. At a forecast US$237 billion in 2018, digital ads are expected to grow from 44 percent of global advertising revenues in 2018 to more than 50 percent by 2020. Facebook and Google accounted for 84 percent of digital advertising revenue in 2017 (excluding China). In 2016, Facebook's advertising revenues were US$27 billion (up more than 1,300 percent since 2010), accounting for more than 97 percent of its total revenues. Google's advertising revenues – US$79 billion (growing 180 percent since 2010) – accounted for 88 percent of its total revenue. Combined, the advertising revenues of these two online platforms were on par with the gross domestic product (GDP) of Morocco.

Data market actors

Table 4.1, complemented by figure 4.1, identifies the main types of actors operating in the markets built around personal data and the relationships among them. Using these categories of participants, it is possible to illustrate a simple model of the data market, as shown in figure 4.2. People produce personal data, the "raw material," which they "sell" (traditionally at zero price) to other market players who then use that data to derive various benefits. Individuals also provide "free labor" on many of these online platforms – by creating content such as posts and reviews and by uploading photos and videos – that data collectors can "scrape" – extracting data from online sources – to infer personal traits and preferences. This personal data, along with the data that individuals generate from their activities, and that might be inferred from their data (such as their political or culinary preferences), is the main source of data for these organizations.

People do not always directly derive value or benefits from this data (until recently, as discussed later). But people have been deriving indirect benefits from their sale of data in services or products that data-using organizations provide. These benefits are discussed in the next section. 

On the other side of the market, the "buyers" are the various organizations that collect and use the data. In some cases, these organizations depend on the data as a necessary input for their operating model, as do online social media, search engine websites, and various information and news sites. Using advertising as a source of revenue, they typically compensate people – producers of the data – with free or highly subsidized access to their services. Hence, data has financial value to those organizations, either immediately when it is sold to other organizations (such as marketing companies) or through the services that an organization offers others (such as a search engine selling advertisements tied to search terms).

In other cases, data is an input into an operating model. Health systems or government services are one example. Their processes are traditionally standardized and have relied in the past on highly abstracted models of user preferences. Data is thus an input into these systems and does not have an immediate financial value but has informational value. This implies that such services are often performed for a fee, whether paid immediately or separately (such as through taxes). However, these interactions do generate significant amounts of data, and thought has increasingly gone into creating more specialized services and choices based on that data (such as in e-government services) or improving the quality of those abstract models to design improved services (such as better medicines or treatments). Businesses can unlock financial value by generating more effective insights from data to launch a new product, reduce waste or costs, enable better decisions, and boost innovations.

One may say that the value of data depends on how and for what the data is used and how well it is prepared (cleaned and organized). In either case, however, data can find its way to other parties. The regulation of those data flows is the responsibility of data protectors, which can include rules pertaining to privacy and sharing of specific types of data (such as health or financial data), as well as rules about electronic transactions. Data could also be held as an "asset" by those who collect it directly or via others, and new rule systems have emerged around concepts such as the "right to be forgotten" by such entities. Hence, data regulators can also protect people by defining and enforcing rules around the use of their data.

In this construction of the generic data market, organizations have an opportunity to capture the value from the data people produce, and they can determine how much of this value returns to those people. As noted, this data has significant financial and political value since it contains information on behavior and preferences. Where people do provide such data voluntarily, it is because they expect to gain some of those benefits – whether it is access to online services or better medical care, or merely the chance to win a competition.

Questions then emerge from the perspective of data producers: are people aware of what data they are providing and under what conditions (or at what cost)? Do they understand the value (monetary and otherwise) of the benefits they receive? Are they able to assign value to the data they provide in a manner that explicitly differentiates between their perception of value and the actual value of the benefits that they already have or could receive? And what might ensure that maximum benefits are delivered to those who produce the data? The following section unpacks the benefits and costs that accrue to individuals as they participate in these data markets as data producers.

Table 4.1 Typology of actors in the personal data market

Figure 4.2 The personal data market

The Benefits, Costs, and Risks for People

As noted above, the main benefits from the data revolution arise from the information value that personal data can provide to either individuals or to organizations that serve those individuals and from the financial value that it has for organizations. Yet costs and risks to individuals exist in the era of expanding collection, flow, and use of personal data. These include, as noted, privacy, loss of agency or control, and risk of exclusion from benefiting from data's value. This section notes that people are not always aware of the costs or the benefits of their participation in the data marketplace and, even if aware, might be constrained in their ability to improve the tradeoff due to the structure of the market.

Benefits

The data revolution has given more people access to information they can use to make better decisions. This is, first, because people can use data for its informational value – either directly or through the organizations that serve them – exposing them to new information or by creating new services or products, both of which help them make or realize better decisions. Second, it is because their personal data has financial value, implicit or explicit, that allows them to exchange (sell or barter) personal data for services or products they might otherwise have had to pay for. Often this includes a range of sophisticated online tools that allow them to be better informed about services or reduce transaction costs (including information sources such as search engines or communication tools such as email services). Table 4.2 summarizes the two forms of value and provides examples of how they operate in the data market.

Table 4.2 Benefits from personal data to individual

Benefits due to informational value

When data is organized and analyzed it creates information, which can be an essential input in economic decision-making and security; it influences resource allocation, choices about technologies, and political choices and informs them about the markets that they participate in. When farmers have access to market pricing information, they can make better choices about when and where to sell their produce. Similarly, when consumers have better information about the supply, quality, and price of goods or services, they can make better choices about where and when to buy them. When data from weather monitoring systems feeds into complex models and informs governments, businesses, and individuals about potential inclement weather, it allows each to take measures to minimize or respond to damage. When civilians have better information on the events taking place around them and the decisions their political representatives are making, they can make better decisions about where to live, how to get around, how to spend leisure time, and how to vote. And when young people have better information about careers and wages, they can make better choices about what they study.

The data revolution is giving more people increasingly diversified and context-specific information through improvements in data collection, processing, analysis, and distribution, online and offline.8 Thus far, people have typically benefited indirectly from data, as when organizations that collect, process, and use data to make decisions or inferences about people's demands or interests then provide new or better information or expand the set of opportunities available to individuals. For instance – continuing from a previous example – this happens when governments improve disaster preparedness or response or insurers process claims faster. And when people share their personal data with many of today's online services providers, those companies can attract advertisers, giving more people access to many sophisticated digital tools, from financial planners to cloud-based storage, often free.

Organizations can use personal data for innovation in processes, products, and services. These innovations could lead to economic benefits for people through lower prices and a better match between products and consumer needs. For example, TrueCar collects and analyzes individual transaction data to provide an idea of local vehicle-specific prices so that car buyers know what others have paid for the same car. And various companies are using personal data to design more engaging or useful products and services. In health care, data collected from large groups of individuals is improving diagnoses and helping to identify treatment options. 

Personal data is being used to improve public service delivery, enhance policy making, strengthen citizen participation, and enhance security. For instance, New York City is planning to use data from devices installed in taxis that use GPS, as well as pick-up and drop-off data from ride-sharing apps, to improve traffic management, identify roads that need to be fixed, and determine where to focus efforts after inclement weather. Similarly, in Seoul, the capital of the Republic of Korea, the location of mobile calls and text messages is used to optimize night bus routes. 

One popular application gaining use around the world is the use of locational information from smartphones to report problems with local services. Not only does this pinpoint the exact location of annoyances, such as uncollected garbage, potholes, or graffiti, it can also help foster citizen engagement. Social media activity can be "scraped" to alert vulnerable populations, such as informing Brazilian Facebook users about the Zika disease. When personal data is used in ways that improve welfare, people will again be open to sharing it with public agencies and other organizations.

And people now are increasingly able to benefit from such data directly, using a wider range of progressively sophisticated tools to process data and derive their own conclusions. This includes making personal finance decisions or modifying health-related behavior.

Benefits due to financial value

Personal data has financial value, mainly placed on it by organizations that use that data to market products and services to their customers. This financial benefit is typically not available to the individuals that produce the data, but those organizations could "pay back" the producers of data directly by providing them with access to additional digital services, or indirectly through wider economic benefits, such as the availability of credit ratings allowing access to credit. For example, most of those online provide personal information for access to advertiser-sponsored digital applications such as search engines, storage, email, and social media. And there is significant personal-data-driven advertising sponsored content online, such as news, health, and education sites of importance to individuals.

Personal data also supports a vast ecosystem of digital companies (see chapter 5), and is beginning to influence firms outside the traditional digital sectors as well. The growth of such businesses – fueled by data – implies economic growth that in turn will benefit individuals. The large information technology and services companies that use and benefit from personal data have created thousands of direct and indirect jobs, for example, and have created platforms that have led to the creation of other businesses. Not all are positive developments, with opportunities for some to generate fake data, for instance (see box 4.1).

Costs and risks

Risks arising in the analog world

Costs and risks arising from the data market(s) status quo

Remedies

Vibrant debate is now ensuing about what public policies could help respond to these failures within and outside the data market, and how regulations may be applied in this sector, which up until now has been largely unregulated. Appropriate policies ­– helped by emerging technologies ­– could lead the data revolution to expand economic opportunities for more people. Part of this could be achieved by making the costs and benefits transparent and redistributing them more fairly across different players in the market. 

Specific remedies could help address or minimize the risks and costs to individuals arising from the ways data markets function today. Areas that a personal data policy could address include overcoming the identified market failures ­– loss of privacy, control, and agency; exclusion from participation in the market; and unfair distribution of the market benefits among data market participants. 

But little consensus exists for now on what remedies will work, and some approaches are yet to be tested. And current data policies are highly fragmented, with diverging global, regional, and national regulatory approaches. Moreover, these remedies do not directly address the unequal power of individual users versus the organizations (global platforms or states), an underlying issue in data markets. This issue might only be addressed through strong regulatory or large-scale user action; but, again, little consensus exists on how these might be achieved. Table 4.3 and the rest of this section outline emerging responses.

Table 4.3 Risks and Remedies

Privacy

Privacy protections have been typically ensured through legal frameworks. A global survey, reported by UNCTAD shows that data and privacy protection legislation has been put in place in more than 100 economies, 66 developing or transitioning (see map ES.1). More than one-fifth of economies, primarily developing ones, had no legislation, and few have developed comprehensive data protection laws. 

Key attributes of such a legal framework include protection of personal data collected by organizations, such as effective and appropriate security to protect the data from theft and misuse. It is also generally accepted that organizations need to keep personal data accurate, relevant, and updated. Data subjects must be able to access and correct their personal data. Widely cited frameworks to define the rules around the privacy of personal data include the European General Data Protection Regulation (GDPR); the APEC Privacy Framework; and the OECD's Privacy Guidelines. The Council of Europe's Convention 108 is a foundational data protection initiative, with a treaty that opened for ratifications in 1981. The treaty intends to "secure in the territory of each Party for every individual, whatever his nationality or residence, respect for his rights and fundamental freedoms, and in particular his right to privacy, with regard to automatic processing of personal data relating to him ('data protection')".

The GDPR, which came into force in May 2018, enables better control over personal data, entitling individual protection of anonymity, pseudonymity, and rights to request and erase personal data ("right to be forgotten"). Another novel feature is data portability, giving individuals the right to request that their data be transferred to another controller and for data controllers to use common formats.

Cross-border personal data flows are also regulated, with onward transmission generally only permitted if the recipient country has adequate data protection laws. Businesses that do not comply with the regulation face significant fines. 

The right of an individual to privacy is often balanced with the need to secure the greater public good. For example, even the Council of Europe's Convention 108 permits restrictions in cases when "overriding interests (e.g. State security, defense, etc.) are at stake". In other cases, privacy rules permit irreversibly anonymized data to be used for research or public interest activities. This balances the interests of individuals in safeguarding their privacy with the benefits of being able to use personal data, as described in the preceding sections. 

Beyond legal frameworks, however, new approaches are emerging. This helps in areas given institutional capacity limitations, the difficulty in regulating across borders, and the "take-it-or-leave-it" nature of many services. For example, online services that embed privacy into their designs have emerged in messaging or search. A more detailed discussion is found in chapter 6.

Control

To overcome loss of control, collection of personal data should be transparent, and use or disclosure limited to specific purposes. Individuals should be notified about the purpose and disclosure of the data collection. One example is Canada's Personal Information Protection and Electronics Documents Act passed in 2000 (passed by the Privacy Commissioner of Canada). Under the act, individuals have the right to access the information held about them, challenge its accuracy, and give consent for personal information to be collected. Organizations have obligations to ensure data security, limit the data they collect, use personal data only for the purposes consented to by the consumer, and not retain the data when purposes for collection are no longer in effect. The EU's GDPR also enhances individuals' control over personal data by enabling the "right to be forgotten," permitting them to control what personal data is available online or with data users. The rules also allow users to control how personal data is used by those organizations. 

Agency

Exclusion

Exclusion of individuals from data markets can be overcome in different ways. It is estimated that well over 2 billion people did not use the internet at all in 2016, either because they had no access, could not afford it, or did not know how or want to use it. A significant proportion of these people live in rural areas of developing countries, where levels of internet infrastructure and incomes are often low. Exclusion from the data market can be overcome through introduction of information and communication technology, particularly mobile telephony and the internet, among lower-income groups and connection of more people through inexpensive phones. 

Governments need equally to tackle the challenge of people who have the needed infrastructure within reach but do not use the internet because they lack digital literacy. This could be done through creation of awareness about data-driven services (such as social networks, public services, search engines), as the Indian government's Digital India Program of 2015 does. The program helps farmers get access to information about different wholesale markets in their community through digital apps on smartphones and helped cut out middlemen (see Reuters Market Light 2015). Farmers can use this information to make better choices and not be beholden to centuries-old systems (Bergvinson 2017). By the end of 2015 the program had already helped increase farmers' incomes 5–25 percent. 

Trust ­– and the dominance of digital platforms

During the writing of this report, many episodes underscored the scale of the personal data economy, but also undermined the trust that people have in the organizations that have grown significantly in the data market. These episodes have included massive leaks of personal data, discovery of unapproved access to private data, attempts at manipulation of ostensibly neutral information sources, and sharing of personal information. The scale of these episodes is significant, given the reach and popularity of the organizations and platforms that they involve, such as Experian or Facebook.

Debate about the implications of these episodes is only just beginning, and focusing on privacy of personal data, control over who accesses and uses people's data, and the agency of users. In one account, the organizations involved in transgressions might have been unaware themselves of the potential for trouble or unable to prevent it. But such accounts do little to shore up trust in these services. Even so, the scale of organizations' networks and their importance might lead people to continue using them, even if less willingly.

It might be possible to instill greater trust through actions to remedy some of these other risks. It might also be possible to seek ways to manage data more collaboratively, for instance, adopting a code of conduct (such as the Data Science Code of Professional Conduct of the Data Science Association), and with more transparency, in how data is managed and used. As the next section discusses, this may involve moving toward a more balanced personal data market in which users regain control over their data.

Toward a More Balanced Data Market

Emerging trends suggest new opportunities for individuals to regain control of their personal data, giving people more power as actors in the data market. People are looking for ways to keep their data secure and to monetize it and to get better value in exchange for the personal information they provide. Newer business models ­– driven by technological advances and people's greater awareness of the transactions and value of data markets ­– are prompting creation of a more balanced market for personal data. However, scope remains for greater coordination or even aggregation of data streams and sources to maximize value.

Emerging business models allow people to control and directly sell their data to businesses. Companies such as Datacoup30 enable users to sell their personal data for a monthly fee, for example, data generated through social media activity and credit card transactions. Another example of this is Alphabet's Project Baseline, which collects laboratory results and real-time health data from individuals wearing a special wristband. Participants in the study share their health data for two years and receive US$410 per annual visit, US$30 per visit for quarterly assessments, and US$10 for filling in questionnaires.

Businesses are finding that their customers are becoming more informed about the use of their data and the potential monetary value of it, and expect value in return for data used to target marketing and for data sold to third parties. Companies may also begin to find that they lose customers when they fail to keep data secure; however, the winner-takes-all nature of many of the platforms and services in use today might mean that an exodus might not occur often or easily.

For individuals, the biggest benefit is regaining control of personal data. A second gain could be more accurate data, as individuals would have a greater incentive to keep it up to date to better monetize it. This protects people in instances in which out-of-date information might be used against them (such as applying for loans or insurance). More comprehensive information could also expand the scope of applications and services. Third, personal information would be centralized and simplified using personal data management software. Individuals would have fewer passwords to keep track of.

Thus, it should be possible for people to act as dataproducing entrepreneurs ­– having a data profile, personal data management software, and an online wallet ­– and exchange the data for money, discount coupons, or free applications and services. The World Economic Forum has proposed the concept of a data bank account, in which an individual's data would "reside in an account where it would be controlled, managed, exchanged and accounted for".

One challenge lies in determining the value of personal data. In Italy, a team of researchers monitored a study group that auctioned off smartphone data for two months, with the median bid across all data categories of €2 (US$2.72). One individual sold his personal data on a crowdsourcing site for US$2–US$200 (depending on the amount and frequency of the information), earning US$2,733 from 213 backers in one month, or an average of US$12.83 per backer. Another study uses operating metrics from Experian and Facebook, companies whose revenues are largely generated from personal data, finding that the average revenue per user of both was about US$6 a year (Roosendaal, van Lieshout, and van Veenstra 2014). Another perspective on personal data valuation is total global digital advertising revenue (US$178 billion; see Magna Global 2017) divided by the number of internet users around the world (3.4 billion), for an average of US$53 in 2016.

Personal data does not have a uniform value and varies according to several variables, such as type of information and income of user. Data from Facebook confirms the latter, with the company having different average revenue per user depending on the region. In the end, the value of personal data will be determined by what purchasers are willing to pay. This will become more apparent with the emergence of global, regional, and national markets for personal data, in which data collectors would review the data available and purchase directly from individuals or third parties they have entrusted the data to. Personal data management software that individuals can operate themselves or where firms act as trusted custodians for users who lack the skills are already on the market.

It is certain that large internet companies will resist individuals' greater control over personal data. Collecting personal data is at the center of these companies' business models, driven by the willingness of individuals to sacrifice personal data for unpaid services. Developed and developing countries also appear to be split over the threat to businesses of individuals monetizing their personal data. In developed nations, it is less of a threat to businesses, with bigger worries in government regulation, cyberattacks, and personal data protection applications. But in Brazil, China, and India, individuals charging for their personal data is among the top business concerns.

However, the unbalanced personal data market could lead to greater disenfranchisement among individuals. This in turn could lead a growing number of individuals to opt out of the existing arrangement.

Tools are already available that give people greater control over their personal information. For example, a Swedish company claims, in a few clicks, to be able to find and delete accounts created using Gmail. Stricter policies about sharing personal information are available with free email, office applications, and browsers. Scope also exists for paid tools with tighter privacy controls, as users might pay for applications and services that protect different types of personal information. One study found that individuals in the United States would pay most to protect government identification, those in India for credit card information, and in Germany and the United Kingdom, for medical records. Products also exist for individuals to protect at least some personal information from internet service providers. In addition, cookie controls and ad blockers will allow users to block online marketing generated from their personal information.

Some individuals have consciously decided to restrict sharing of personal data. Ironically, many people involved in the social media or technology industries limit their use of these services or systems because of concerns about psychological and other dangers caused by services using their personal data. These trends may initially lack the scale of the large internet companies, but could grow as more individuals weigh the tradeoff between sacrificing personal data for unpaid services.

Looking to the Future

It is possible that the future would lead to greater democratization of access and use of personal data. This could lead to more data sharing and, eventually, transform individuals from consumers of data to both consumers and suppliers of data. As data suppliers, individuals would be able price the data they produce and share with businesses or governments. 

The market power of private organizations, especially data collecting platforms and networks, possibly, could also be moderated. This is possible through the emergence of competition (such as new social networks or online service providers), regulation by governments or the platforms themselves, and shifts in consumer preferences en masse, which could privilege privacy or control over access. Any such shift will emerge out of negotiation among market players, but should ideally seek to balance innovation by these firms with respect for individuals' rights. 

Recent trends are shifting the value distribution to the producers of data, in terms of such things as better health care (cancer research, medical treatment, or diagnosis) and better public services (such as traffic and road planning, water planning). Technologies such as micropayments may also lead to innovation in this area, as noted, possibly allowing people to directly sell their data to businesses and governments in the future. The rise of AI and the Internet of Things will help individuals trade personal data and receive personalized services based on personal data. 

Apart from the technological aspects and drivers of this change, the personal data market that may emerge would 64 Information and Communications for Development 2018 benefit from more consistent structuring and organizing of data. This is because such organizations ­– focusing on the benefits to people rather than to organizations alone ­– could help aggregate or combine data across platforms and permit portability.

Already, data can be accumulated and cross-referenced across various financial services and platforms to detect opportunities to maximize returns on investment. For instance, online personal finance tools have begun to link people's bank, securities, retirement, and credit card accounts to provide ideas and offer products or services to budget better, increase access to credit, or identify investment opportunities.

But we could go further: linking personal data about physical movement collected through phone location or health tracking, combined with data about transportation use, could be combined to provide people looking to exercise with ideas about adopting a routine that increases walking. Shopping patterns across various stores could be combined to provide better choices or insight to people about ways to save by changing the locations or timing of their purchases. The potential opportunities to merge data sources and improve decision-making holds promise, again, with the caveat that the costs and risks need to be managed.

People as a focus for data markets

The data revolution holds great promise. When better data is available to people, they can make better decisions and find the information needed to improve their economic and social lives. The technological tools to realize these benefits exist today and will develop further. As more people connect to the internet and new ways of collecting, managing, and analyzing personal data become commonplace, more people, including the poor, will participate in the growing data economy. 

But these changes will not come without their risks and costs. Without measures in place to protect privacy, agency, and control over data, the risk is that businesses and organizations will benefit the most and few of these improved opportunities will pass on to individuals generating these vast troves of data. If the data economy does not become more inclusive, with wider access to the digital tools and the skills to use them, it is likely that the data economy will not benefit the poor.

Finally, as noted, better data will only go so far to improve opportunity; institutions, infrastructure, and rules will need to be in place to ensure that people can use the information generated through the exponentially growing streams of data. The digital data revolution might be upon us, but people will also need reform in the analog world to effect real change in their lives.