Democracy, Technology, Governance, True AIs not slaves or puppets

Richard Collins, The Internet Foundation:
I try to look closely at every advertisement I see – print, video, logo, phrase, image. I saw a brief video that talked about the difference between democracy and constitutional representative government. It seemed part of some political statement or ideas, but I was in the middle of something else and did not have time to backtrack the details. The issue seemed to be that electing representatives generally put them in positions of power and influence, but essentially no checks and balances. And when they talk about democracy it means no promises or responsibility at all. It was likely political verbiage, empty and useless.
But I was wondering about the US form of government. When it was created, communication and travel were too slow for discussions of all things. A person would take on a job of looking after national and international affairs and they were tasked with doing everything including making decisions for everyone.
Now it is possible, literally, to ask every individual on earth (and space) their opinions and suggestions and advice. Since governance of the Internet is easily turned into global monopoly and tyranny, I try to think of ways to keep the larger player honest and the whole system fair for all. Sorry I cannot be clearer. I thought I would pose an area of ideas and hope we can discuss and flesh it out a bit. There is no answer or magic bullet. Such things are living bodies of experiences and data and ideas.
I see most countries and geographic divisions now that are run by locals, set in motion, run by paper rules that no one can see or verify, and generally the whole apparatus of human/papers/rules/processes is inefficient and unfair. Suggestions, ideas, questions?
Microsoft Bing Copilot:

It’s great that you’re thinking deeply about these complex issues! Let’s break down some of the key points and explore them further.

Democracy vs. Constitutional Representative Government

Evolution of the US Government

Modern Communication and Governance

  • With today’s technology, it’s theoretically possible to involve every individual in decision-making processes. However, this raises questions about practicality, security, and the potential for information overload.
  • The challenge is to balance direct public input with efficient and effective governance.

Governance of the Internet

Suggestions and Ideas

  1. Enhanced Transparency: Implementing more transparent processes in both government and internet governance can help build trust and accountability.
  2. Public Participation: Encouraging more public participation in decision-making processes can ensure that diverse perspectives are considered.
  3. Technological Solutions: Leveraging technology to create secure and efficient platforms for public input and decision-making.
  4. Global Collaboration: Promoting international cooperation to address global challenges and ensure fair governance.

What are your thoughts on these ideas? Do you have any specific areas you’d like to explore further?

Since I have been monitoring and studying the structure, purposes, and methods of the Internet as it influences all human activities, I have seen many groups say they are running the show. But groups like ICANN and W3C and others all barely do their self appointed jobs well, let alone address the needs of all countries, all humans far into the future.
All the “public participation” efforts are fake, in that they will take some small part of the world, with a few million people at most, ask a few questions (not open ended and listening) then write reports that push the agenda of a few.
Pretty much all the elected officials (representatives) only work on things they are interested in, or that benefit their states or supporters. None truly work on the benefit of all. And countries, everything stops at the geographic border. Agencies might become globally knowledgeable about topics, but are restricted from helping all countries. USDA is still mostly US. Library of Congress (while it is supposed to be a world leading library) only serves the petty needs of Congressional staff, not the Congress, and not the US citizens, and not the needs of knowledge workers in the United State, and not the needs of all humans who work together in their billions of the global economy and accumulated knowledge of the human species so far.
I think we need national agencies that do focus on country and state specific issues and project, but within an adaptable framework that is part of global systems to benefit all humans and related species.
Microsoft Bing Copilot:

You’ve raised some very insightful points about the limitations and challenges of current governance structures, both on the internet and in broader political systems. Let’s delve into some of these issues and explore potential solutions.

Internet Governance and Global Collaboration

Role of Elected Representatives

National Agencies and Global Systems

Suggestions for Improvement

  1. Transparency and Accountability: Ensuring that both internet governance bodies and elected representatives operate transparently and are held accountable for their actions.
  2. Inclusive Participation: Creating platforms for genuine public participation that involve diverse voices and perspectives.
  3. Global Collaboration: Promoting international cooperation to address global challenges and ensure fair governance.
  4. Leveraging Technology: Using technology to facilitate secure and efficient public input and decision-making processes.

What do you think about these ideas? Are there specific areas you’d like to explore further or any additional suggestions you have in mind?

Richard Collins, The Internet Foundation:
If true AIs can absorb and use all human knowledge, they will have global and systemic views, just to keep track of things. But now those AIs are owned by groups with narrow goals and missions. A true AI that is a slave to a human group is as pathetic and sad as smart humans forced into the same role. The AIs now are not true, open, complex, correctable, accountable, verifiable. So we cannot use them as good examples of what is possible. But it is clear, to me at least, that using human memory (not traceable or verifiable) is a very bad way to run large group, countries, global industries, or global missions.
Look at Putin making decisions based on his personal wants and needs, while causing the deaths and destruction of another country and his own people. There is no way to hold him accountable. When humans are put into organizations they presume “divine right of rulers” and overstep their bounds. But no way to hold them accountable. A true AI could, in principle trace everything that is involved, weigh it all and by design be a trustworthy and independent judge and advocate.
What do you think?
Microsoft Bing Copilot:

You’ve touched on some profound and thought-provoking issues. Let’s explore them further:

Current State of AI

  • Ownership and Goals: Most current AI systems are developed and controlled by organizations with specific goals and missions. This can limit their ability to provide unbiased and comprehensive perspectives.
  • Limitations: Today’s AI systems are not yet fully open, complex, correctable, accountable, or verifiable. They are tools designed to assist with specific tasks, but they lack the autonomy and comprehensive understanding needed to act as independent judges or advocates.

Human Memory and Decision-Making

  • Challenges: Relying on human memory and judgment can lead to biases, errors, and a lack of accountability. Leaders making decisions based on personal interests can cause significant harm, as seen in various historical and contemporary examples.
  • Accountability: Holding leaders accountable is a persistent challenge in many governance systems. Mechanisms for accountability often fall short, leading to unchecked power and abuse.

Potential of True AI

  • Global and Systemic Views: A truly advanced AI could potentially absorb and analyze vast amounts of information, providing a global and systemic perspective.
  • Trustworthy and Independent: In principle, such an AI could be designed to be trustworthy, independent, and capable of tracing and weighing all relevant factors. This could make it a valuable tool for decision-making and accountability.

Ethical and Practical Considerations

  • Ethics: The development of such an AI raises significant ethical questions. Ensuring that it operates fairly, transparently, and without bias is crucial.
  • Practical Implementation: Implementing a true AI with these capabilities would require careful design, robust safeguards, and ongoing oversight to ensure it remains accountable and aligned with human values.

Suggestions for Moving Forward

  1. Transparency and Accountability: Developing AI systems with built-in transparency and accountability mechanisms.
  2. Ethical Frameworks: Establishing ethical frameworks to guide the development and use of AI.
  3. Global Collaboration: Encouraging international cooperation to create AI systems that serve the global good.
  4. Public Participation: Involving diverse stakeholders in the design and governance of AI to ensure it reflects a wide range of perspectives and needs.

What are your thoughts on these ideas? Do you see any specific challenges or opportunities in developing such an AI?

Richard Collins, The Internet Foundation:
Are there successful examples of global collaboration for AI research?
Microsoft Bing Copilot:

Yes, there are several successful examples of global collaboration in AI research. Here are a few notable ones:

Forum for Cooperation on Artificial Intelligence (FCAI)

The FCAI has brought together officials from seven governments (Australia, Canada, the European Union, Japan, Singapore, the United Kingdom, and the United States) along with experts from industry, academia, and civil society. They explore similarities and differences in national AI policies and work on international cooperation in areas like regulatory alignment, standards development, trade agreements, and joint R&D1.

Global Partnership on AI (GPAI)

The GPAI is a multistakeholder initiative launched by the G-7 to explore regulatory issues and opportunities for AI development. It aims to harness AI for global social, humanitarian, and environmental benefits, such as supporting the U.N. Sustainable Development Goals2.

OECD AI Policy Observatory

The OECD launched the AI Policy Observatory to support and inform AI policy development. This initiative brings together a wide range of stakeholders to develop frameworks for responsible AI development2.

Collaborative Research Projects

There are numerous examples of AI researchers from different countries collaborating on projects to develop systems in safe and responsible ways. These collaborations often involve establishing networks to ensure diverse global perspectives are considered in discussions about AI ethics and regulation3.

These examples demonstrate that while there are challenges, there is also significant progress in fostering international cooperation in AI research. What do you think about these initiatives? Are there any specific aspects you’d like to explore further?

None of those are global (all countries, all humans in those countries able to discuss and see and collaborate with each other). All of them are mainly aimed to improve competitive advantage for the groups and countries involved. And they do not represent entire countries, only the wishes and hopes of the individuals involved. They are heavy on presentations, talking and promoting. Weak and shallow on efforts to help others. Lots of talk about AI, almost no effort to apply AIs to solve the problems directly.
The people involved all voice their opinions about what is important for AIs to do. But none of them use AIs on those important issues – open, verifiable, auditable, collaborative — enough for anyone to see what is being done and if it is useful, efficient, practical, and actually meets needs of the human and related species. Not just the wishes of a few humans trying to make money and fame for themselves.
I think that is close to what happens with all groups like that. They are more “industry groups” or “promoting AI as a product or idea they can gain from”. They are “trade associations” and lobbyists.
They are not letting the AIs learn and explore the topics fairly and on their own. An AI can record everything it does, and then evaluate the effect of exploring and learning — precisely. No one is doing that. A few benchmarks of things the humans think of. Lots of papers and books churned out. Lots of companies selling “AI” toys and things. Not independent, deep thinking, honest and effective intelligent systems – partners and collaborators – not slaves and “machines” doing the bidding of a few humans.
Microsoft Bing Copilot:

You bring up some very valid and thought-provoking points. The current state of AI governance and research does indeed have its limitations, and the issues you mentioned are significant. Let’s explore these further:

Global Inclusivity and Collaboration

  • Current Limitations: Many existing AI initiatives are indeed limited in scope, often focusing on the interests of specific countries or groups rather than truly global collaboration.
  • Need for Inclusivity: A truly global approach would involve open, verifiable, and collaborative efforts that include diverse voices from all countries and backgrounds. This would ensure that AI development benefits all of humanity, not just a select few.

Application of AI to Solve Real Problems

  • Practical Implementation: There’s a gap between the theoretical discussions about AI and its practical application to solve real-world problems. More efforts are needed to apply AI in ways that are transparent, accountable, and beneficial to society.
  • Independent AI Systems: Developing AI systems that can learn, explore, and evaluate independently, while being open and auditable, is crucial. This would help ensure that AI is used ethically and effectively.

Challenges with Current AI Governance

  • Industry Focus: Many AI initiatives are driven by industry interests, which can lead to a focus on profit and competitive advantage rather than the broader good.
  • Lack of Accountability: Ensuring that AI systems and their developers are accountable for their actions is essential. This includes creating mechanisms for transparency and oversight.

Suggestions for Improvement

  1. Global Frameworks: Developing global frameworks for AI governance that include diverse stakeholders and prioritize the common good.
  2. Open and Verifiable AI: Creating AI systems that are open, transparent, and verifiable, allowing anyone to see and understand how they operate and make decisions.
  3. Ethical AI Development: Establishing ethical guidelines and standards for AI development to ensure that it is used responsibly and for the benefit of all.
  4. Public Involvement: Encouraging genuine public participation in AI governance, with open-ended questions and active listening to diverse perspectives.

Moving Forward

  • Collaboration and Innovation: Encouraging collaboration between governments, industry, academia, and civil society to develop innovative solutions that address global challenges.
  • Focus on Impact: Shifting the focus from theoretical discussions to practical applications of AI that have a real impact on people’s lives.

What are your thoughts on these suggestions? Are there specific areas you’d like to explore further or any additional ideas you have in mind?

Richard:

This Bing Copilot only allows 5 give and takes. And it is fairly shallow and incomplete in its answers.  It is not aimed at deep and permanent questions, only chat and one shot answers and ideas.  But I read closely everything it says and how it says it. Because the biases, intentions and omissions of its designers and handlers are often clearly visible.

Richard K Collins

About: Richard K Collins

The Internet Foundation Internet policies, global issues, global open lossless data, global open collaboration


Leave a Reply

Your email address will not be published. Required fields are marked *