IAS Project

Explore

Notes

(500 words) The first section will introduce the idea of values as central to the design process of algorithmic systems (whether knowingly or unknowingly). It will also introduce the idea of socio-economic incentives as the underlying selector for value picking.

Values and the design process

Values are always part of the design process of algorithmic systems. Algorithmic systems are the ever-changing bounds of what we refer to as artificial intelligence. Values cannot be avoided when private-sector technology companies design an algorithmic system nor when private-sector technology companies design platforms utilising such algorithmic systems. Thus, we shall restrict our scope to private-sector designed platforms, which are themselves composed of algorithmic systems.

For proprietary algorithmic intellectual property, we see ex-post ethical issues arising within all stages of the design process: design, implementation, and use.

Insert examples of what kinds of issues arise in all stage of the design process.

Design: value extraction model.

Implementation: biased datasets

Use: barriers to entry and economic outcomes.

AI designers have been successful in focusing the larger innovation ecosystem on technologies that focus on wealth extraction (capital accumulation).

Idealistically, private-sector platform designers governance structures should be centering around a common good and public value as their bottom-line, not capital accumulation.

Linking sentence to: the fake garden of Eden.

How can we shift some of the ‘design process’ power legally granted to proprietary AI firms, towards a more distributed decision-making process that includes other interested parties?

Socio-economic incentives beyond those that benefit private sector AI designers directly are mere acts of benevolence, and are not structurally embedded.

So, we look into the relationship between private-sector platform designers incentives for the design process and their tendency toward wealth-extraction.

⁠

Traditional proprietary dynamics and the space between.

Proprietary IP and open-source

This section might not be needed. Its about the economic incentives of corporate governance. open-source is a choice made by corporate governance. Otherwise it would be a policy recommendation about open-source.

This section will sketch a continuum stretching from proprietary intellectual property to open-source, seeing how conceptions of property and governance change on this continuum.

There is one way to interpret the socio-economic spectrum for proprietary dynamics of technology platforms. On one end of the spectrum technology platforms are locked-private, whilst on the other they are locked-open. Locking things private is an appeal to libertarianism, where private property and proprietary approaches to property are embraced. An example of a technology company taking advantage of this is Microsoft, which achieves high levels of wealth extraction from its proprietary IP with marginal regulatory bounds. On the other end, platforms that are locked-open appeal to anarchist ideals of property, where things are open and non-proprietary. Wikipedia is an example of a locked-open platform, whose use-value is free to use and extractable by any internet user.

There is one main issue at each end. The issue with locking things private is that rewards resulting from proprietary IP act as a strong incentive structure and thus capital accumulation concentrates to its relevant parties. Whilst the issue with locking things open is that volunteering is a comparatively weak incentive structure (how do you get paid?) for development and its use-value can be utilised by locked private organisations and contribute without retribution to the aforementioned issue of locking things private (loss of use-value by resourceful private firms).

Linking sentence.

Open source could be a starting point, but not an end. Exploring the space between the extremes of proprietary IP and open-source can help devise a socio-economic incentive structure that aims for public value in a relatively proprietary environment. What kind of property and governance structures can facilitate this?

⁠

Wealth extraction

We must acknowledge that there is a distinction between income derived from adding value and wealth gathered through extraction. The first is a reward for taking chances that increase an economy's capacity for production; the second results from receiving an excessive share of the reward without making equivalent enhancements to the economy's

productive capacity⁠

. Wealth extraction is the process by which value is extracted from a system and transferred to a small group of individuals or entities. In the context of the digital economy, wealth extraction refers to the process by which value is extracted from digital platforms and transferred to the platform owners, often at the expense of users, workers, and other stakeholders.

(insert real world examples of this: Google and Microsoft/OpenAI)

⁠

Establishing effective public monitoring should ensure public value creation through digitalisation and AI (

AI for the common interest⁠

The rewards of value creation in the (AI-powered) platform economy must be shared more equitably using new digital architectures (The Value of Everything).

⁠

There are also regulations governing the rights of shareholders, such as requirements for disclosure and transparency, restrictions on insider trading, and protections against fraud and abuse.

⁠

DAO-enabled stakeholder governance affordances

DAO-enabled binding stakeholder governance addresses several specific issues with shareholder governance, including:

Centralisation: shareholder governance concentrates decision making power in the hands of its wealthy shareholders.

Decentralisation: DAO-enabled stakeholder governance, on the other hand, distributes decision making power among all stakeholders part of the DAO.

This minimises the configuration of AI-systems towards capital accumulation for shareholders.

This minimises the reliance on the benevolence of the few, instead establishing a dynamic of reciprocal

What about decentralisation?

“Decentralised and distributed modes of organisation are well defined in computer science discourses and denote a particular network topology. Even there, they can be understood either as an engineering principle, a design aim, or an aspirational claim. In the decentralisation discourse these three dimensions are often conflated without merit. A decentralised network design might not produce decentralising effects and might not either necessarily be decentralised in its actual deployment.”

^This is relevant to me.

“When the technical decentralisation discourse starts to include social, political, or economic dimensions, the risk of confusion may be even larger, and the potential harms of mistaking a distributed system for something it is not, even more dangerous. Individual autonomy, the reduction of power asymmetries, the elimination of market monopolies, direct involvement in decision making, solidarity among members of voluntary associations are eternal human ambitions. It is unclear whether such aims can now suddenly be achieved by particular engineering solutions.”

^ Need to prove this in my piece.

“An uncritical view on decentralisation as an omnipotent organisational template may crowd out alternative approaches to creating resilient, trustworthy, equitable, fault resistant technical, social, political or economic modes of organisation.”

^ Precisely, it is about distributing decision-making power to more relevant parties. This is justified in circular economic theory.

lit review

There is considerable literature about the potentials of blockchain for the governance of digital commons. Rosaz et al. (2021; 2021) identified a set of affordances of blockchain technologies for achieving commons governance. The research maps such affordances to ‘Ostrom’s 8 Principles for Managing a Commons’.

However, this does not itself structurally disrupt the logic of accumulation undertaken by AI system designers. So, our focus will be more specific: how DAO affordances can disrupt the shareholder driven logic of accumulation by powering stakeholder governance.

There has been

Similarly, we are interested in the affordances granted for achieving binding stakeholder governance.

Leveraging blockchain for the non-profit sector to enhance good governance:

⁠

https://www.frontiersin.org/articles/10.3389/fbloc.2022.1083647/full⁠

⁠

Blockchain for carbon off-sets markets. “The power dynamics changed because the use of blockchain for Veridium’s carbon offset market caused additional stakeholders to join the network of stakeholders.”

⁠

https://www.frontiersin.org/articles/10.3389/fbloc.2022.957316/full⁠

⁠

There are often conditions of ‘decentralisation’ that must be satisfied for the above to be a DAO, but what kind of decentralisation and how is it measured?

“The governance mechanisms we identify (Mechanisms of Governance) are setting proposals, voting, tokens, sanctions, reputation or participation records, constitution, consensus, and validating. Governance mechanisms are used to maintain, update, and upgrade the ecosystem and the ledger itself. Although mechanisms are essential for implementation, more research is needed to determine which mechanisms provide conditions for a viable decentralization of ecosystems.”

“According to the research data, a constitution is an essential mechanism to formalize governance and create trustworthiness for the ecosystem. An example of this process is the constitutionality assessment in Tezos. However, if all the rules of governance are assumed to be coded as smart contracts, ecosystems still have open questions concerning the need for arbitration mechanisms. The research data shows that ecosystems, such as IXO, are prepared to offer mechanisms to resolve disputes between stakeholders through formally defined arbitration processes.”

“According to

Beck et al. (2019)⁠

, the central role of incentives is to be a new dimension between the traditional digital economy and the blockchain economy. According to our data, incentives are considered in some ecosystems. However, there are no signs that incentive alignment could be accomplished without decent governance mechanisms, such as proposing, voting, and arbitration. Thus, although economists may see incentives as omnipotent determinants of behavior, governance instruments are also needed to align and regulate the behavior of stakeholders in decentralized ecosystems.”

⁠

https://www.frontiersin.org/articles/10.3389/fbloc.2021.613115/full#B2⁠

⁠

https://aisel.aisnet.org/cgi/viewcontent.cgi?article=1835&context=jais⁠

⁠

of AI corporations to non-profit, public organisations - for example, the European Commission’s High-level expert group on Artificial Intelligence - would give configuration power to a body that is incentivised to embed interests beyond those of capital accumulation.

⁠

What I need to show in this paragraph is how governance affordances allocated in DAOs address the issues of corporate benevolence and capital accumulation. Of course, both of these issues are addressed via the decentralisation that results. The kind of decentralisation here is that of xxx...

Stakeholder governance emphasises the importance of transparency, accountability, and stakeholder participation in decision-making processes [16].

Still aims for capital accumulation.

A corporation with stakeholder governance may still prioritise profits, but it seeks to balance the interests of multiple stakeholders in the pursuit of that goal, again, without being legally or structurally bound to do so.

Decision rights of stakeholder governance are ambiguous.

In a corporation with stakeholder governance, the governance structure is designed to give voice to multiple stakeholders, but the decision-making process may still resemble that of shareholder governance.

What this distinction highlights is that DAO governance rights must avoid self-monitoring, self-reward, and self-punishment (Moldoveanu & Martin, 2001).

If DAOs issue purchasable fungible or non-fungible governance tokens, akin to shares, in theory they become no different to shareholder governance. On the other hand, if the DAO issues non-purchasable fungible or non-fungible governance tokens, an alternative method of distribution must be determined.

Returning to our discussion about shareholder models of governance for AI-system designers, we advocate for AI corporations to utilise DAOs to be the infrastructure for their corporation to implement a stakeholder model of governance.

⁠

The first affordance to be allocated is what rights each type of stakeholder member has in the DAO. The inclusion of stakeholders to the DAO does not necessitate equal rights to all DAO members. This time, the configuration must be done on the DAO itself to establish what internal governance will look like. If the goal is to grant some degree of rights to actual stakeholders, then a way to do so (either legally or strategically) must be established.

Whilst on the other hand, we identify two

The granularity of what the on-chain governance mechanism can do depends on its technical structure. How the members are selected and the power of each member to contribute to the governance mechanisms mentioned are flexible features that must be decided by the members of the DAO.

Now that we know what a DAO can be, the aim is to show how DAO-enabled binding stakeholder governance models help prioritise stakeholder value over shareholder value for AI corporations. To do this, we must show how the governance process of an AI-system designer could be run on-chain.

Affordance of DAO governance

Rights: decision management vs decision control.

Accountability: technical vs institutional accountability.

ex-ante accountability.

“In the blockchain economy, accountability will increasingly be enacted technically instead of institutionally, in principle at least.”

Incentives: consensus mechanisms vs governance mechanisms (development, maintenance and use).

DAO membership mechanism

But how much power should stakeholders have in DAOs?

But what should the governance structure look like?

This is a topic of research beyond this piece. There has been development in quadratic voting and so on.

DAO affordances

This is because of three affordances granted by DAOs:

transparency of design of product and governance. Doesnt need to be all open-source, but everything around it must be transparent.

ex-ante method of in-house governance.

complex stakeholder table of governance (with less decrease in decision-making speed compared to cooperatives).

⁠

This⁠

guy says that regulators should “consider ways to cooperate with engineering communities to develop codes that appropriately facilitate mechanisms to achieve regulatory goals”

In traditional shareholder governance, decisions are made by the board of directors, who are accountable to the shareholders. In stakeholder governance, decisions are made by a wider range of stakeholders, such as employees, customers, suppliers, and the local community, who are all affected by the decisions made by the organization.

By using a DAO, stakeholders can have a more direct say in the decision-making processes, which can help ensure that decisions are made in a fair and transparent way that takes into account the needs and interests of all stakeholders, rather than just the shareholders.

Because of such affordances, DAO-enabled stakeholder governance models would obstruct AI-system designers from pursuing whatever means to satisfy their logic of accumulation. OpenAI should not have an easy time transitioning to pursue such ends. It currently does, and there is nothing stopping it.

While both governance of the infrastructure and governance within the infrastructure are relevant, the assumption is that governance of the infrastructure works, and governance within the infrastructure addresses our issue.

The aim is to show that a legally protected DAO infrastructure for private sector AI system designers

We explored blockchain as an artifact to facilitate the governance of AI system designers, with an emphasis on shareholder corporations.

Lack of Transparency: shareholder governance does not necessitate transparent decision making

Transparency: DAO-enabled stakeholder governance, on the other hand, operates through transparent smart contracts, ensuring that all members have access to the same information and can see how decisions are being made. The transparency of DAO proposals can be limited to stakeholder visibility. This minimises

Short-Term Thinking: Shareholder governance can incentivize decision-makers to prioritize short-term profits over long-term sustainability. DAO-enabled stakeholder governance can prioritize long-term sustainability by including rules and mechanisms that ensure the protection of the environment and the interests of future generations.

Limited Accountability: Shareholder governance can sometimes lack accountability, with decision-makers not being held responsible for their actions. DAO-enabled stakeholder governance, on the other hand, includes mechanisms for holding decision-makers accountable through transparent voting and decision-making processes.

Exclusion of Stakeholders: Shareholder governance can sometimes exclude stakeholders such as employees, customers, and local communities from decision-making. DAO-enabled stakeholder governance can include these stakeholders as members of the DAO, giving them a say in decision-making and ensuring that their interests are represented.

What is a DAO? How would it be implemented in the current corporate governance structure of the private sector, specifically for AI companies?

⁠

Conclusion/abstract sentences

Decision making power about how the DAO progresses is distributed to its members.

⁠

Stakeholder model

Should consider the interests of each stakeholder in its governance process.

Stakeholders are considered as external to the governance process.

Stakeholders vary depending on the purpose and output of the company, but examples include: -

creditors

auditors

customers

suppliers

government agencies

community and public

Notes

A change from the shareholder model to a stakeholder model that includes a deeper understanding of public value generation will be necessary to develop a new economic basis that disrupts the logic of accumulation.

It is important to recognise that the public, private, and civic spheres collaborate to create wealth and other good market results.

Both the logic of AI technologies and of the overarching AI innovation ecosystem, freely shaped by capital-accumulation-based business models, were the incubators for today’s AI companies. Therefore, we must tackle the problem as encompassing the logic of accumulation as non-confounding factors. Developing a different kind of private sector governance model will require regulating governance structures to redirect the incentives of AI-system designers away from the logic of accumulation.

DAO-enabled must-stakeholder model

why

Must consider the interests of each stakeholder in its governance process.

Stakeholders are considered as internal to the governance process.

The normative advancement of ‘must’ here is to force the consideration of the interests of the stake holders by adding them to the governance process.

The assumption is that democratising the governance process reduces the risk of benevolently acting in the interest of the stakeholders. The aim is to act democratically, not benevolently.

how

Putting organisations within a DAO digital infrastructure enables

Reimagining the AI platform economy: from shareholder to extending stakeholder power.

The current form of the platform economy: identifying the socio-economic problem.

(500 words) This section will ground the philosophical analysis from section 1 to the real world, looking at the roots of the AI-enabled platform economy and the socio-economic issues with value extraction.

Economic incentives and decision-making

Let’s centre the socio-economic incentives of the AI system designers and their effect on the whole design process.

For proprietary AI firms in a capitalist economy, incentives beyond those that benefit the firm are a mere act of benevolence. Therefore, the internalised values of the algorithmic systems of such firms - at the design, implementation and use level - are decided (explicitly or implicitly) by self-serving, market driven incentives.

Dominating AI technology platforms have been successful in focusing the larger innovation ecosystem on technologies that focus on wealth extraction. This is evident from the findings of Zuboffs Surveillance Capitalism.

Wealth extraction and big tech

This claim remains agnostic as to what is considered a common good and what would count as public value, but whatever they may be, there is evidence that we are currently feeling both ends of the stick simultaneously: the emergence of monopolistic platform designers such as Google, Amazon and Airbnb and the ‘public value’ they benevolently bring.

Idealistically, private-sector platform designers governance structures should be centering around a common good and public value as their bottom-line, not capital accumulation.

Pragmatically, some international organisations advocate a particular type of ethical framework for general private-sector algorithmic system designers.

Common Based Peer Production (CBPP) of open-source

This subsubsection will look at what Common Based Peer Production (CBPP) is and its 6 principles for governance.

Linking sentence to: the fake garden of Eden.

How can we shift some of the ‘design process’ power legally granted to proprietary AI firms, towards a more distributed decision-making process that includes other interested parties?

Digital Policies & Recommendations: the fake garden of eden.

Not enough: Justified by Donut economics lady.

Technology enabled digital political economies: digital commons and new governance structures.

(500 words) This section will introduce contemporary digital commons governance theories that aim to democratise the design process of technology firms.

Stakeholder Theory

Partial Common Ownership (PCO)

Plural Voting and Ownership (PVO)

Why not CBPP?

Ostrom has shown that there are certain properties that emerge from such collaboration. But how is this put into practice? PCO shows this practice through quadratic voting.

Name

Link

Read

Comments

To Do

Priority

Reimagining the Platform Economy

From shareholder to stakeholder

A change from the shareholder model to a stakeholder model that includes a deeper understanding of public value generation will be necessary to develop a new economic basis. It is important to recognize that the public, private, and civic spheres collaborate to create wealth and other good market results. It is no longer sufficient to base business decision-making and policy analysis exclusively on concerns with maximizing efficiency (cite).

Both the logic of AI technologies and of the overarching AI innovation ecosystem, both freely shaped by capital-accumulation-based business models, were the incubators for today’s AI-enabled platforms. Therefore, we must tackle the problem as encompassing both value extraction and market power as non-confounding factors. Developing a different kind of AI economy will require regulating market dynamics to shape both the aim of the technologies so as to eliminate value extraction and minimise market power.

For example, imagine an economy where public commons procurement and common market standards altered the shareholder model to one of stakeholder partial common ownership for private sector. Under this proposed dispensation, what would be effected is the internal governance model and incentive structure of all compliant AI-enabled technology platforms.

The gained compensation here is not about renumeration, but the broader effects on the innovation ecosystem and the aims of its constituents, if users were not only the top priority, but embedded int he innovation cycle. This predication is not anti-Microsoft, OpenAI, Google or Amazon. Rather, it is arguing against the current prima-facie templates of governance and incentives that enable private actors to act against the interest of its users through extraction, privacy violations, and all other known anti-social practices.

Wikipedia and open-source - Radicalxchange pdcast

“How can we shift some of the power that is concentrating in large tech firms, towards a more distributed outcome among the general public?”. Talked about this by looking at the relationship between Wikipedia and lots of tech we use today and the value of the wiki infrastructure to all these other products.

RETHINKING ART OWNERSHIP: PARTIAL COMMON OWNERSHIP AS A STEP TOWARDS A MORE SYMBIOTIC ECOSYSTEM

"[DLTs] have created a space for deconstructing and reconfiguring what ownership means and how it is practised. Instead of a monolithic exclusive property right, ownership can be radically reimagined, making it possible to positively entangle the interests of multiple parties in context-specific and unprecedented ways.”

If we approach algorithmic platform ‘ownership’ as an institution for the ecosystemic support of value creation between its various actors, then we must think beyond platform value-extraction.

Suppose we consider four kinds of interests: the communities and industries from which algorithm platform design emerges, the designers themselves, owners of the algorithmic platforms designed, and users. The ties between these interests are established by conventional ownership, but new forms of ownership may completely reimagine them.

Conventional ownership empowers the owner of the platform – whether that’s a private organisation or individual through IP rights – giving them control over the value of the symbolic link between the user and the community/issue.

“What if, instead, the platform could function as a fluid social or financial link between the user and the source community/issue? Might a different model of ownership better facilitate that type of engagement?”

“We believe there are ways of opening up ownership that would better recognise and respect the social embeddedness and collective/networked nature of cultural creation. One example is Partial Common Ownership (PCO), – a hybrid form of ownership that levels the playing field between incumbent art owners and others; encourages more liquidity; and maintains alignment between the interests of collectors, creators, and communities.”

PCO opens up ownership to better recognise and respect the social embeddedness and networked nature of platform value-extraction.

Between Scarcity and Abundance

The Value of Everything

A Deeper Investigation of the Importance of Wikipedia Links to the Success of Search Engines

Mission Economy - Guide to changing capitalism

Partial Common Ownership

⁠

Github⁠

⁠

ChatGPT:

Partial Common Ownership (PCO) is a type of ownership model that seeks to balance the interests of different stakeholders involved in cultural production, such as collectors, creators, and communities. In PCO, ownership is divided into multiple partial shares, with different parties holding different portions of the ownership. This allows for greater participation and collaboration among a wider group of individuals or entities, rather than being limited to a small group of traditional owners. Additionally, PCO creates a more fluid and flexible ownership structure that enables easier transfer of ownership and encourages greater liquidity. Overall, the goal of PCO is to promote a more equitable and inclusive ownership structure that better reflects the social embeddedness and collective nature of cultural creation.

Plural Property

Plural Property is about partial common ownership of property. Can this be IP too, and with firms?

Plural Voting

Use this on a DAO structure to make decisions?

There are no rows in this table

⁠

DLTs for the sharing economy: realising digital commons theories through DAOs.

(500 words) This section will look at how DAO research and implementations are able to realise digital commons theories

(200 words) Why DLTs and not blockchain.

Realising Stakeholder PCO through DLT enabled DAOs for the platform economy.

what are Stakeholder Plural Property models of governance for digital platform economies?

achieve a socio-economic incentive structure in a relatively locked-open environment through partial common ownership of proprietary IP.

DLTs and DAOs

This subsection will give a brief introduction to Decentralised Ledger technologies (DLTs), Decentralised Autonomous Organisations (DAOs), and their relevant affordances: tokenisation, self-enforcement and formalisations of rules, autonomous automisation, decentralisation of power over the infrastructure, transparentisation, codification of trust.

DAOs have been analysed to offer the following affordances: tokenisation, self-enforcement and formalisations of rules, autonomous automisation, decentralisation of power over the infrastructure, transparentisation, and codification of trust.

Partial Common Ownership.

This section will give a brief introduction to Partial Common Ownership (aka Plural Property, COST or SALSA). Partial Common Ownership (PCO) opens up ownership to better recognise and respect the stakeholder embeddedness and networked nature of platform value creation. The aim is to see how changing the conception of IP ownership in platform economies can address the issue of wealth extracting incentives.

paragraph...

If we approach platform ‘ownership’ as an institution for the ecosystemic support of value creation between its various actors, then we must think beyond platform value-extraction. Suppose we consider four kinds of interests: the communities and industries from which algorithm platform design emerges, the platform designers, owners of the algorithmic platforms designed, and the users of the platform. The ties between these interests are established by conventional proprietary ownership, but new forms of ownership can completely reimagine these ties. Conventional ownership empowers the owner of the platform – whether that’s a private organisation or individual through IP rights – giving them control over the value of the symbolic link between the user and the community/issue.

Suppose the platform could operate as a flexible social or financial connection between the user and the source community or issue. In that case, would an ownership model that includes the source community or issue be more effective in promoting that type of interaction? There are methods for expanding ownership that would more accurately acknowledge and honour the social connectedness and collaborative, networked qualities of cultural production. Partial Common Ownership (PCO) is one such instance, which is a blended ownership model that balances the power dynamics between existing platform owners and new participants, encourages greater flexibility, and sustains harmony among the interests of platform users, designers, and stakeholders.

Here is how it PCO works. The license terms require the license holder to evaluate and announce the value of the license every two years and put it up for auction on the public market. Additionally, the license holder must pay a fee that is proportional to their declared value (higher declared value equals a higher fee). The critical component that makes this system work is that if another party bids more for the license, it is awarded to them. This encourages license owners to avoid declaring a low price, as they could lose their right to the platform to a competing bidder. Conversely, they can make it difficult for other bidders to surpass them by setting a high price, as long as they can compensate stakeholders appropriately through the proportional fee. It is important to note that if a license holder is outbid, it does not harm them because they still receive the higher bid amount. As the platform's value grows, the entire network, including users, its designers, and its stakeholders, share the value gain equally.

Want to print your doc?
This is not the way.

Try clicking the ⋯ next to your doc name or using a keyboard shortcut (

CtrlP

) instead.