9 Democratizing the AI Economy

Since the global financial crisis, numerous debates and discussions have taken place on how to reform the market capitalist system of free trade, globalized financial markets, and privately owned corporations so that it respects the preservation of the natural environment and the welfare of local communities, promotes social equality, and produces not only profits for the few but also social goods for the many. More recently, the Business Roundtable, a business group of more than 200 members, including the chief executives of JPMorgan Chase, Amazon, and General Motors, and of a combined annual revenue of US$7 trillion, published a statement of purpose calling for a significant departure from the dogma that businesses should serve only the owners of capital.1 Companies should “protect the environment” and treat workers with “dignity and respect,” the statement said. Without doubt there is growing awareness in corporate America and elsewhere that Milton Friedman’s philosophy of businesses caring only for maximizing value for their shareholders is outdated. We now live in a different world. The threat to life on our planet due to climate change requires corporate leaders to think long term and not just of the next quarterly results. Millennials and younger generations are more sensitive to issues of corporate responsibility and sustainability and often boycott brands seen to be all about healthy balance sheets and little else.2 Moreover, there is considerable pressure for reforming corporate purpose from politicians, especially as corporate profits soar and many companies opt to spend those profits on share buybacks instead of improving worker wages.

And yet, there are limits to what governments in liberal democracies could do. They could certainly regulate executive pay more and rebalance tax systems so that corporate profits are taxed more while work is taxed less. They could introduce smart policies for dealing with technological disruption, for instance, by using tax incentives to promote technologies that generate new and more high-value work for humans rather than simply automating tasks, and by future-proofing health care, pension, and welfare systems by taking into account the impact of automation technologies on workers. Such measures and policies would surely smooth the impact of intelligent machines in our societies over the next few years and would quite possibly reestablish some degree of citizen trust in democratic institutions.

Nevertheless, as I have argued throughout this book, we also need to think beyond what governments can traditionally do to regulate markets, and explore more radical approaches to reforming capitalism in the Fourth Industrial Revolution. We need to think out of the box in order to deal with the incoming disruption, both because of the enormity and novelty of the challenges that await us and also because of the unique opportunities provided by intelligent technologies. In this chapter I would like to focus on two such opportunities. The first is how we can go several levels deeper from the macroeconomics of capitalism and reimagine some of the fundamental elements in the machinery of corporations. More specifically, I will explore how the governance of digital platforms can be redesigned in order to give birth to new types of business organizations that can have purposes and goals beyond profitability. Second, I will look into the “new oil” of the AI economy—namely, data—and discuss how we can create new, game-changing institutions and data-driven paradigms that could transform every person in the world into a value-creating asset.

Reinventing Business Governance

Companies are not only economic entities but political entities too; their organizational model and method of governance reflect the zeitgeist of their times. As the first industrial revolution took hold in Victorian Britain, joint-stock companies acquired the principles and language of representative democracy. Shareholders were often called “constituents.”3 Like a parliament, general assemblies of shareholders would elect directors—the “executive government”—and collectively decide on important issues using a voting system based on the number of shares. According to an observer in 1807, the General Court of the East India Company was akin to a “popular senate,” with the important difference that voting rights did not discriminate between the sexes.4 Robert Lowe, a nineteenth-century British statesman, called these companies “little republics.”5 The direct participation of shareholders in a company’s governance began to diminish as company operations expanded around the globe and shares started trading over the counter of stock equity exchanges. Thus, the little republics of nineteenth century early industrial capitalism6 regressed to dictatorships run by omnipotent boards of directors who ruled over strictly hierarchical organizations. As stock markets grew in size and importance, shareholders could now buy and sell shares of any publicly listed company, anytime, and so they lost interest in the purpose of the companies they invested in. Shareholder interests also diverged from the interests of employees, suppliers, customers, and local communities where those companies were based. The long termism and plurality of goals that characterized early joint-stock companies was gradually replaced by the short termism and the singular purpose of quarterly financial performance.7

The advent of the Internet in the late twentieth century set off the transformation of the global economy into a “digital economy” and, as discussed earlier, enabled a new business model to emerge, the digital platform. Google, Microsoft, Apple, and Amazon are companies harvesting the value that is created and added by participants interacting and transacting on their digital platforms. They are also the most valuable companies in the world today based on market capitalization.8 But the digital economy did not disrupt corporate governance. Digital companies have retained the characteristics of “enlightened dictatorships” run by visionary CEOs and boards of directors. Centralization of decision-making is still at the heart of corporate governance in digital, as well as less digital, companies today. Without doubt, centralization confers important operational efficiencies, but it can also promote groupthink and hinder agility, innovation, and resilience to change. What concerns us here is how the current, centralized governance model in most corporations exacerbates income and wealth asymmetries, particularly in light of advancing automation due to AI—and whether there is an alternative.

As discussed in the previous chapter, web 2.0 platforms are profoundly different from web 3.0 cryptoplatforms. The former are essentially centralized rent-extraction machines, while the latter can be decentralized and widely distributive of the economic value they generate. Figure 9.1 summarizes the main differences between the two business paradigms and compares them to Industrial Era hierarchies.

Web 3.0 cryptonetworks9 can be peer-to-peer networks that resemble “mutual organizations,” such as cooperatives or mutual insurance companies, in that members derive rights to profits through their participation as workers as well as customers.10 Cryptoeconomics provides the theoretical framework and blockchain the technology to deliver a more democratic governance of digital platforms on web 3.0. Participants in such platforms may earn tokens for their participation, use those tokens to exchange economic value, and also store them as an asset of value that grows as more exchanges take place across the network. Thus, the network effect that results from participants’ contribution returns some of the value created directly back to the participants. Because tokens are lines of code, there is virtually no limit to the degree of innovation that can occur with respect to ways of incentivizing participants on web 3.0 cryptoplatforms.

Tokens can also be used for voting on matters of governance. For instance, in proof-of-stake protocols, participant voting power is proportionate to the amount of tokens each participant has. This approach has power balance consequences when a small majority may end up controlling more than 51% of stakes and imposing its will on the rest. In such cases the minority may disagree with the way that the platform is being developed or managed. In a centralized governance world, a minority of shareholders would have a very limited set of options to counteract an authoritarian majority; they can protest, go to the courts if they can afford it, or sell their stock and bear the losses. But in a decentralized cryptonetwork, a disenfranchised minority can do much more: they can make a hard fork and create a new blockchain. Forking, by being the “nuclear option” of dissenting minorities on a web 3.0 platform, provides the ultimate inhibitor to a controlling majority becoming authoritarian, for if they did, they would run the risk of many valuable participants in the network simply walking away and becoming competitors, which would then lower the financial value of the network.11 Forking is therefore a mechanism for realigning the interests of all participants, regardless of the varying degrees of voting power that they may have.

The crucial interplay between the balance of political power on a cryptoplatform and the economic value that the platform can produce requires a layered approach to decentralized governance (see figure 9.2). At the most basic layer, the founding stakeholders must agree on a set of fundamental principles, a kind of “constitution” for the platform. It is at this constitutional layer that human interaction and deliberation can make use of the citizen assemblies’ model. Founders may come together from a range of stakeholder groups, such as workers, software developers, hedge funds, foundations, local governments, entrepreneurs, and so on. Those founders will have to first agree on the purpose of the platform, its business objectives, and its values and ideals; for example, they may decide to safeguard jobs, offer health insurance, pensions, or childcare, or protect natural resources. The constitutional layer must also be compliant with national and local laws and regulations. By democratizing the process of defining the foundational constitution, a web 3.0 cryptoplatform can be established as one of those “little republics” that Robert Lowe was describing in the nineteenth century—or indeed as a Cyber Republic of the twenty-first century!

Once the foundational constitution has been decided on, software developers can start building the second, derivative layer of the cryptoplatform, where smart contracts will track, audit, and regulate exchanges on the basis of what the constitution allows and promotes. This would be the operating system layer of the cryptoplatform, where the mechanisms of recording of transactions and tokenization will be built. On top of this layer, other developers will then build the distributed applications that will run on the platform and produce economic value for everyone, thus creating the value-creation layer. Finally, an additional supervisory layer is needed to resolve issues that are certain to occur as the cryptoplatform evolves over time. For example, at this layer, stakeholders will be able to address and vote on resolving technical issues and conflicts, as well as making complex decisions. Although the foundational principles of the cryptoplatform will have been decided in the cryptoplatform’s constitution, the supervisory layer must also allow, in extreme cases, the community of stakeholders to alter aspects of the constitution and propagate these changes to the other layers as well.

Organizations of the Future

This participatory, layered, collaborative model of governance and codevelopment can be used to imagine new types of business organizations. Figure 9.3 shows how business ecosystems on web 3.0 that are decentralized and democratic in their governance could function as a “stack” of services that may themselves be delivered by networked cryptoplatforms. These ecosystems would connect a variety of businesses including cooperatives, traditional corporations, or other “microbusinesses” (e.g., small teams of professionals).

In such a future scenario, web 3.0 ecosystems may connect to other ecosystems, too, and thus create a fractal economy of interconnectedness and cooperation facilitated by networks of open blockchains. For example, take a hypothetical open cryptoplatform that provides a car-sharing service. This “full stack” cryptoplatform is a self-contained ecosystem of software developers, designers, drivers, riders, social media experts, communicators, marketers, and so on, governed by its constitution and its smart contracts. This autonomous ecosystem, however, is dependent on the existence of infrastructures such as roads and the provision of fuel,12 as well as the provision of other services such as car insurance, car maintenance, car mobility qualification, and so forth. All those ancillary services can be provided by other cryptoplatforms. Moreover, many public services, such as public utilities and road maintenance, can also be reimagined in the context of cryptogovernance, as we shall see in more detail in the next chapter.13

Cryptoplatforms can be autonomous and interconnected at the same time. And anyone can be a participant, and contribute in various roles, in more than one cryptoplatform. We can thus imagine organizations of the future that use blockchain to track the contribution of millions of participants, a type of “swarming” organization that delivers some social good. The example of Wikipedia is perhaps one way of visualizing those multimillion participant businesses of the future, the significant difference being that in those future businesses participants will also be remunerated for their contributions via tokens.

Cryptoplatforms can also be completely leaderless, as the Decentralized Autonomous Organization (DAO) on the Ethereum blockchain demonstrated in 2016.14 The DAO aimed to showcase the feasibility of a revolutionary form of business organization that had no owners and run without the need of executive management or a board of directors. It was crowdfunded via the issue of a token and managed to raise over US$150 million from around 11,000 investors. The business purpose of the DAO was to invest in various ventures and projects and distribute the profits from those investments back to the token holders. Flaws in the computer code resulted in the DAO getting hacked, and millions of investor dollars getting siphoned away into the hackers’ accounts. Ethereum ultimately managed to recover the funds, but only after making the very difficult decision to “hard fork” the blockchain. As a result of that forking, the Ethereum blockchain comes today in two “flavors,” Ethereum and Ethereum Classic. Nevertheless, the DAO remains a model for future organizations that needs to be improved, but also regulated. For instance, the US Securities and Exchange Commission (SEC) has decreed that digital tokens, such as the ones traded in DAO, should be considered securities and therefore fall under the respective laws that require the registration of their owners, including compliance checks on money laundering and terrorist activities.15 Such regulation is essential if those new types of organizations are ever to leave the experimentation stage and enter the mainstream.

A Sharing Economy for Workers

Web 3.0 cryptoplatforms can also allow us to imagine gig-working platforms that serve the interests of workers and not only the interests of platform owners or investors. This is of vital importance for the future, as automation is likely to result not in the complete elimination of work but in the gradual replacement of steady, salaried jobs with part-time and intermittent employment.

To return to the ride-sharing example, take Uber, currently valued in the tens of billions of dollars. There is no reason why a group of drivers shouldn’t be able to build a similar platform for themselves: the software tools to develop it are open-source and readily available, and development and design talent can be given incentives to participate. A blockchain-based ride-sharing platform called La’Zooz,16 founded by Shay Zluf in Tel Aviv, was one such proof of concept. The blockchain protocol allowed anyone to join and begin to earn Zooz tokens each time they drove over 20 km (about 12.4 miles), or contributed code to the design of the app, or got others to join. Zooz tokens would then be used as the internal currency of the ride marketplace. Similar initiatives have sprung up elsewhere too. In Singapore, a Korean team called MVL introduced TADA, the equivalent of Uber on blockchain. The designers of TADA17 view their application as a means to create a new ecosystem that would include drivers, riders, and automotive industry companies and adjacent service industries and their customers. In China, a company called Didi18 is building another blockchain-based ride-sharing application called VV Go, which they plan to test in Singapore and then extend to other cities, too, such as Toronto, Hong Kong, and San Francisco.

Another example of using blockchain to reinvent the gig economy is Supp,19 an application that connects contingent hospitality workers and those who wish to hire them. The market that Supp addresses from their headquarters in Melbourne is rather sizable; it is estimated that around 800 million families globally earn their living in the hospitality industry,20 while according to the UN around 200 million millennials are traveling around the world, working on and off in hospitality and leisure jobs. This market is currently fragmented and subject to the web 2.0 platform model exploitation. As Jordan Murray, cofounder of Supp, explains,21 “We believe the gig economy is essential to the future work mix, but the corporate model used by Uber and Airbnb leads to poor outcomes for workers and users who are the backbone of these platforms. Through decentralisation we plan to make a platform where there is no distinction between network owners and network participants. The concept is called a ‘decentralised autonomous co-operative’ and our MVP has already created more than AUS$170,000 in shifts for its members in Melbourne.”

Supp connects workers to hirers just like any gig worker marketplace. Payment of wages is made in fiat currency; however, a token is automatically deposited into a worker’s “tip jar” upon completion of a shift. Token economics running as smart contracts on Supp’s blockchain not only ensure that the added value of participation returns to the participants but also reflect the shared principles of the community and thus deliver many positive externalities aside from pure commerce. Supp is a “decentralised autonomous co-operative” because of its democratic governance. It members own tokens that give them rights to vote (“one token one vote”). Supp’s governance model has in-built provisions written in its smart contracts that encourage good behavior. For instance, a worker stakes a token on a shift upon accepting, and in the event of no-show the token is forfeited to the hirer. Tokens unlock governance rights and have increasing utility over time. As their value increases, tokens can also be exchanged for fiat money over cryptocurrency exchanges.

A particularly interesting characteristic of Supp’s use of blockchain is how it handles user data. In web 2.0 platforms, user data are siloed and owned by the platform. Arguably, one of the reasons why these platforms are so exploitative is because of personal data “lock-in”: workers cannot move their data to another platform, and indeed the cost of transition becomes higher over time. Rent-extraction machines on web 2.0 succeed because workers are vested in the platform and have little choice but to agree to onerous terms and conditions. By contrast, web 3.0 cryptoplatforms such as Supp offer self-sovereign identity and full data ownership rights. Supp users own their data on the blockchain and can take their data with them anytime they wish and join another platform. These data include their personal data as well as their reputation data. By pioneering data self-sovereignty in practice Sapp is pointing to a future where citizens are not powerless spectators but active participants in shaping the destiny of their countries and nations.

Data Self-Sovereignty and Data Trusts

Data property rights are an essential element for solving the wealth asymmetries of the Fourth Industrial Revolution.22 In the AI economy algorithms are the pipes that deliver business value, and data are the fuel, the “new oil.” There are several categories of data that AI systems need in order to train. What concerns us here are data generated by citizens. There are at least two classes of citizen data. The first comprises those that are strictly personal, such as data about one’s age, sex, body measurements, achievements and failures, preferences and dislikes, health and employment records, and so on—let’s call this class “personal data.”23 A second class of citizen data comes from social interactions and economic transactions, for example, data created each moment we interact via a digital channel with other people, groups of people, companies, institutions, courts, governments, and so on—let’s refer to this class as “social data.” Not all data have equal value, and most data only have value when they aggregate with other data. For example, the RAND Corporation calculated that in order to get accurate AI models for self-driving cars, there is a need to collect data from 500 billion to 1 trillion miles driven.24 Unless individual drivers start sharing their driving data, building deployable models for self-driving cars may prove to be an impossible task for any company. Finally, there are data over which citizens must have complete control with respect to who has access,25 and other data over which they must have partial control or none at all, for example, their medical or criminal records. Today, most citizen data are owned either by governments or by private corporations. This puts citizens at an economic disadvantage since they have no stake in the economic value that their personal and social data generate.

Regulation of personal data is currently restricted to privacy only, and is mostly centralized. The European Union’s GDPR is an example of centralized governance of personal citizen data. Although the regulation is aiming to protect user privacy, it falls short from granting full property rights of the data to their rightful owners. Arguably, any centralized arbitration over data rights is subject to the intrinsic problems of representational governance: power asymmetries will always result in the strongest and the richest having more influence over the final result of arbitration compared with the weakest and the poorest. The principal-agent problem will emerge every time a centralized authority makes decisions on behalf of the many. Regardless of how “enlightened” the EU leadership may be, or how sincerely they may care for the data rights of European citizens, web 2.0 technologies will always favor the owners of the platforms more than the data-producing users.

One way of solving the problem of redistributing more widely the value created by our own personal data is to set up “data trusts.” This is an idea pioneered by the Open Data Institute in the United Kingdom.26 Data trusts can be mutual organizations with fiduciary responsibilities that act as stewards of citizen data. Citizens sign up to data trusts and pool their data together in order to make them available to businesses, governments, or other organizations that may be interested in using them. Data trusts negotiate with those interested organizations and offer access to the data in exchange for fees or a share of Intellectual Property (IP) rights derived by their use in training and powering AI algorithms and applications. Data trusts can be organized in various ways—for example, similarly to traditional pension trusts where a board of trustees has fiduciary responsibility for how the trust invests and distributes wealth. Such representational forms of governance do suffer, as discussed, from the principal-agent problem, which could be overcome by implementing data trusts on a blockchain and using cryptoeconomics to democratize its governance. Decentralized data trusts based on cryptogovernance models can be set up by cities, NGOs, private companies, worker unions, or grassroots citizen movements. They solve the principal-agent problem of democratic governance while accelerating innovation in AI by unlocking economic value in citizen data. They could also provide a framework to protect property rights for citizen data and thus contribute to reducing wealth asymmetries in the AI economy of the future. Finally, they could provide a defense against both corporate and state surveillance. As long as citizens are in control of their data, their privacy is more secure.

Decentralizing AI

Today, several problems hinder the wider adoption of AI by small business and society at large. They range from lack of a clear strategy and shortage of specialized skills—such as adequate numbers of AI engineers and data scientists—to shortage of good and varied data sets. Eager to regulate, governments are adding onerous compliance costs on top of all that, making it even harder for small businesses that aspire to ride the growth wave of AI. Thus, regulation and scarcity of skills and data combine to favor the big tech giants that can use their dominant position and economics of scale to advance AI ahead of everyone else. In other words, the AI economy is not a level playing field but is massively influenced by an oligopoly of players. Is there a way to change the rules of the game?

The open-source movement has shown how peer-to-peer networks can accelerate innovation by creating communities that share values and knowledge, as well as code. Cryptoeconomics can take this proven paradigm to a whole new level by adding economic incentives and democratic governance. By putting those two things together, we can imagine a “decentralized AI marketplace,” an exchange for AI tools, data, and applications built using blockchain technology, where network effects and cryptoeconomics accelerate innovation and lower costs, while significantly lowering the barriers for participating in the AI economy. To deliver such a marketplace, one needs three principal building blocks. The first block is a data marketplace, where anyone can provide his or her data securely in an open data exchange. Ocean Protocol27 is one of the most prominent innovators in this space. By creating a cryptonetwork for data, they are delivering a way for citizens, communities, and business to share data on a marketplace securely and transparently while keeping ownership. Data trusts could also be major contributors in the supply side of such decentralized data marketplaces.

The second building block for a decentralized AI marketplace is decentralizing cloud computing so that AI innovators can train models in a cost-efficient way, without lock-in to the cloud of big tech providers. Currently, just four centralized cloud computing providers—Microsoft, Amazon, IBM, and Google—own more than 70% of the cloud computing market.28 This oligopoly prevents prices from dropping to a level where anyone, anywhere in the world, can become an innovator and create economic value. The cost for access to computing will become an even more critical factor that will decide the wealth asymmetries of the future. As we enter the Fourth Industrial Revolution, deep learning networks will be one of the most proliferative technologies: they will power image and language recognition systems as well as a host of other applications embedded across business systems and infrastructures.29 These technologies are computationally intensive, and this is one reason why the current cloud computing oligopoly is at odds with the interest of society at large: it raises cost-to-access barriers and skews wealth creation toward capital owners and against ideas creators and data owners. The UK-based start-up DADI30 is building a decentralized cloud computing marketplace, with a vision to provide a viable and preferable alternative to the current cloud computing oligopoly. DADI uses blockchain to create a peer-to-peer network where anyone can connect his or her computing device and earn income by providing spare computational capacity for rent. Decentralized cloud computing services, such as DADI, are challenging the status quo.

The third building block is a decentralized marketplace for AI models, applications, and tools; that includes cost-efficient frameworks where innovators can create, assemble, and market their ideas. A decentralized AI marketplace ecosystem that interconnects data, computing, and development frameworks is a hub for fast innovation and democratic economics.

Here’s how the whole thing would work: developers solve for real business problems and deliver innovative AI applications to businesses, organizations, and communities. They do so by collaborating with end customers over the AI marketplace in order to define the problem that needs solving. They then deliver the solution by accessing, combining, and evolving more elementary AI tools and models, which are supplied by the wider AI developer community on the platform. Since data are the fuel for successful AI applications, these AI innovators access unique data sets that are supplied through the decentralized data exchanges and data trusts. Training and deploying the models takes place using the decentralized computing infrastructure, which lowers the cost of innovation and systems use. Using cryptoeconomics, tokenized rewards flow across the wider ecosystem, and each participant on the supply side (AI tools and applications, data, computing, ideas, etc.) is rewarded according to the participant’s contribution. This creates powerful incentives for all participants. Artificial intelligence innovators can monetize their work, and businesses, communities, and individuals can turn their business data into profitable assets. Customers get what they need at a very low price. And there is no lock-in to a centralized cloud provider. Everybody wins.

The Big Transformation

Artificial intelligence is a general-purpose technology that will be the main driving force of the innovation and wealth creation that will take place in the Fourth Industrial Revolution. In combination with blockchain-powered cryptoplatforms, the dividends from the emerging AI economy can be distributed more fairly and widely and thus help reduce the wealth and income asymmetries that are currently fueling popular mistrust in free markets and the institutions of liberal democracy. In the years to come, many other technologies will enter the “plateau of productivity” with profound impacts on our world. One could imagine, for example, the tremendous impact of decentralizing energy production using ultraefficient solar panels at every home, and what that would mean for localizing energy distribution grids and reducing transmission costs and environmental degradation.31 Or, the significance of decentralizing manufacturing using 3D-printing technologies;32 or, perhaps, the revolutionizing of food production using a combination of urban farming and genetic engineering so that everyone can produce most of their food closer to home.33 In all those cases citizens will become producers and consumers of their fundamental needs at the same time. We have witnessed a similar transformation during the first phase of the information revolution, with the advent of social media, whereby citizens became both the producers as well as the consumers of content. Web 3.0 technologies have the potential to extend this transformation by including just about everything and can thus create new markets of millions of “prosumers” where political power is inherently decentralized.

In such a truly interconnected peer-to-peer society citizens can own tokens across many web 3.0 platforms. Smart financial systems and exchanges can help them manage and invest or trade their tokens, so that they can absorb troughs when income might be scarce, as well as pay for the provision of essential services in health care, insurance, education, and retirement. Interconnected ecosystems of democratically governed cryptoplatforms that power AI systems can reinvent the “sharing economy” by distributing the creation of new wealth among their token holders, while rewarding fairly those who contribute more to the collective welfare. The digital tokenization of these new peer-to-peer markets could thus reform capitalism and make it work for the many and not just the few.34

Moreover, these new peer-to-peer markets can transform how we think of ourselves as workers and active members of society. In traditional capitalism workers are a potential liability, for they must produce more than they consume in order to have any value. In a tokenized web 3.0 economy workers become value-producing assets by virtue of their existence, as value always feeds back to them from their individual inputs. Collectively, this would mean, in effect, the true realization of an autopoietic cybernetic system incorporating humans and their technologies at a societal scale. For, whether it will be our data, or our other intellectual or physical contributions, we would be providing via our joining, participation, and interactions with the various peer-to-peer cryptoplatforms the raw energy that could power the engines of sustainable and resilient economic well-being in the future.