TL;DR:
- Concerns about the potential harm to democracy from AI systems like ChatGPT and GPT-4 due to the volume of autogenerated arguments that could potentially mislead individuals.
- Risk of these AI systems being deployed by businesses for market dominance, putting democracy at risk.
- Potential for AI systems to be used for the greater good and bolster democracy through a publicly-focused AI system developed by the government and made available to all citizens.
- Cutting-edge AI systems are currently controlled by tech giants who dictate how we interact with them and determine our access, putting democracy at risk due to their corporate interests.
- Need for AI systems that serve as public goods and prioritize the public good rather than being controlled by tech monopolies.
- Existing LLMs trained on internet-sourced material may reflect bias and hate, with companies rushing subpar products to market.
- Lack of democratic oversight in the decisions made by these companies that impact democracy.
- Benefits of a public option Large Language Model (LLM) as a crucial independent source of information and a testing ground for technological choices with significant democratic consequences.
- Concerns about LLMs “hallucinating” and inferring false facts, undermining democracy.
- The potential of LLMs is to aid citizens in formulating and expressing their perspectives, testing new applications that support democracy, serving as explainers and educators, and facilitating radical democracy.
- Importance of robust AI for democracy, with the need to capture human feedback for specific democratic use cases.
- The cost of training the largest-scale LLMs is substantial but manageable, with government labs, universities, and nonprofits with AI expertise and capability able to oversee the development and deployment of a public option AI system.
- Public option AI system would ensure that the interests of the public are put first, rather than being controlled by tech companies with their own agendas, improving the formulation and expression of perspectives, educating citizens, facilitating civic debates, and helping groups reach consensus.
Main AI News:
The Rise of AI and its Impact on Democracy: An Analysis
Artificial Intelligence (AI) systems, such as ChatGPT and GPT-4, have raised concerns about their potential harm to democracy. The sheer volume of autogenerated arguments produced by these systems could potentially inundate public discourse and sway individuals down misleading paths. These fears are not unfounded, as businesses seek to deploy untested AI systems in a bid for market dominance, thereby putting democracy at risk.
However, there is another future that could unfold – one where AI systems are used for the greater good and bolster democracy. This requires an AI system not controlled by tech monopolies but rather developed by the government and made available to all citizens. Such a public option is attainable if desired.
A publicly-focused AI system would be custom-made for use cases where technology can best support democracy. It could aid in citizen education, facilitate collective deliberation, summarize public opinion, and identify common ground. Politicians could leverage large language models (LLMs) like GPT-4 to better understand the needs and wants of their constituents.
Today, cutting-edge AI systems are controlled by tech giants such as Google, Meta, and OpenAI in partnership with Microsoft. These companies dictate how we interact with their AI systems and determine our level of access. As they are motivated by corporate interests, they could shape AI systems to align with those interests, thereby threatening democracy.
It is imperative that we seek AI systems that serve as public goods and prioritize the public good. Existing LLMs are trained on internet-sourced material, which may reflect bias and hate. While companies try to filter and fine-tune their LLMs to eliminate these biases, leaked conversations and emails suggest that they are rushing subpar products to market in a bid for monopoly.
These companies make critical decisions that impact democracy yet face little democratic oversight. It is unclear if LLM-powered chatbots and search engines favor certain viewpoints over others or avoid controversial topics. The public must trust these companies to be transparent about the trade-offs they face, but it is imperative that we seek AI systems that operate with democratic oversight.
The Benefits of a Public Option Large Language Model (LLM)
A public option LLM could serve as a crucial independent source of information and a testing ground for technological choices with significant democratic consequences. Similar to public option healthcare plans, a public option LLM would increase access to information while promoting transparency and putting pressure on private products to improve pricing and features. It would also allow us to understand the limitations of LLMs and direct their applications accordingly.
The potential for LLMs to “hallucinate” or infer false facts is a significant concern. Trusting technologies that generate false information could undermine democracy, and companies selling these technologies cannot be relied upon to admit their flaws.
However, a public option AI system could do much more than just verify the honesty of technology companies. It could test new applications that support democracy rather than undermining it.
LLMs could aid citizens in formulating and expressing their perspectives and policy positions, making political arguments more informed and coherent. This does not mean that AI will replace human participation in political debate, but rather, it will assist citizens in articulating their thoughts and beliefs. The use of AI for speechwriting, lobbying, and campaign messaging is already being tested and speculated upon. A public option LLM could serve a similar role for everyday citizens as professional speechwriters serve influential individuals.
If the “hallucination” problem can be solved, LLMs could also serve as explainers and educators. Citizens would be able to query an LLM with expert-level knowledge on policy issues or political positions, gaining a deeper understanding of political matters. LLMs would be unfailingly available and endlessly patient, providing an unprecedented resource for political education.
The Power of AI in Facilitating Radical Democracy
Artificial Intelligence (AI) has the potential to revolutionize the way we participate in democratic processes. As Carnegie Mellon Professor of Statistics Cosma Shalizi has noted, we delegate decisions to elected politicians in part due to a lack of time to deliberate on every issue. AI has the ability to manage large-scale political conversations and identify common positions, surface compelling arguments, and minimize hostile or disruptive behavior.
AI chatbots could run national electronic town hall meetings, automatically summarizing the perspectives of participants, providing a dynamic alternative to opinion polling. In the future, AI could even help groups reach a consensus and make decisions. Experiments by AI company DeepMind suggest that LLMs can bring people with differing opinions to a common ground.
However, realizing this future requires a public option AI system. Building such a system would require significant effort and political will, but the potential benefits to democracy are substantial. Many AI tools are already publicly available, with tech giants like Google and Meta cooperating with the academic community to make their latest tools freely accessible.
Financially, the cost of training the largest-scale LLMs is substantial but manageable, estimated at hundreds of millions of dollars. This cost is a fraction of US federal military spending and a wise investment for the potential return. Government labs, universities, and nonprofits with AI expertise and capability could oversee the development and deployment of a public option AI system.
The Importance of Robust AI for Democracy
The release of poorly tested AI systems poses a threat to democracy. To avoid this outcome, it is crucial that AI systems are robust and designed to strengthen, rather than undermine, democracy.
Recent advancements in AI chatbots were made possible through feedback from real people, who taught the models which outputs were useful and which were not. However, these models were aligned with corporate interests, not democratic values. To build AI that supports democracy, it is necessary to capture human feedback specifically for democratic use cases, such as moderating policy discussions, explaining legal proposals, or articulating perspectives in a larger debate. This will ensure that the AI aligns with democratic values and that mistakes can be corrected without damaging users and the public arena.
A public option AI, developed by the government and available to all citizens, could be tailor-made to support democratic processes. This AI could educate citizens, facilitate deliberation, summarize perspectives, and find common ground. Politicians could use LLMs to better understand the views of their constituents.
The Importance of a Public Option AI for Democracy
The current state of AI systems is dominated by large tech companies, who have the power to shape AIs to conform to their corporate interests. This poses a threat to democracy, as these companies make decisions with huge consequences for democracy but with little democratic oversight.
A public option AI would provide a vital independent source of information and a testing ground for technological choices with significant democratic consequences. This would increase transparency in the sector and put productive pressure on private products. Additionally, it would allow for a better understanding of the limits of AI and the direction in which its applications should be directed.
LLMs often “hallucinate,” inferring false facts. This raises concerns about citizens trusting technologies that make up random information and the companies’ honesty in admitting their flaws. A public option AI could not only check technology companies’ honesty but also test new applications that could support democracy.
Looking further into the future, these technologies could help groups reach a consensus and make decisions. Early experiments by the AI company DeepMind suggest that LLMs can build bridges between people who disagree, helping bring them to a consensus. Science fiction writer Ruthanna Emrys, in her remarkable novel A Half-Built Garden, imagines how AI might help people have better conversations and make better decisions—rather than taking advantage of these biases to maximize profits.
However, it’s important to note that this future requires responsible and ethical development and deployment of AI systems. The risks of AI systems being misused to harm democracy are real, and it’s crucial that we consider these risks and find ways to mitigate them as we explore the potential of AI to support democracy.
An AI public option could help ensure that these systems are developed and used in a way that benefits society as a whole rather than serving the interests of a few powerful tech companies. By making AI systems accessible to all citizens and by incorporating feedback from the public in their development and deployment, we can ensure that AI systems are aligned with democratic values and are used to support the public good.
To build assistive AI for democracy, it is necessary to capture human feedback for specific democratic use cases, such as moderating a polarized policy discussion, explaining the nuance of a legal proposal, or articulating one’s perspective within a larger debate. This allows for the alignment of LLMs with democratic values by having the models generate answers to questions, make mistakes, and learn from the responses of human users without having these mistakes harm users and the public arena.
A public option AI system would ensure that the interests of the public are put first rather than being controlled by tech companies with their own agendas. This public option AI system could be developed by government agencies or universities and nonprofits with AI expertise rather than being controlled by tech monopolies.
The development of a public option AI system would require a significant effort, both technologically and politically. However, the potential benefits for democracy, such as improving the formulation and expression of perspectives, educating citizens, facilitating civic debates, and helping groups reach consensus, make it a worthwhile endeavor.
Conlcusion:
The development of a public option AI system that is designed to assist democracy could have a profound impact on how citizens engage with the political process. From helping citizens express their perspectives to serving as an educator and explainer of policy issues to facilitating large-scale political conversations, the potential benefits of such a system are vast.
However, the development of such a system will require significant effort, both in terms of technological development and political will. It is crucial that we approach this development with caution and consideration, carefully testing and experimenting with smaller, more constrained use cases before scaling up. By doing so, we can ensure that AI will serve as a tool for democracy rather than a threat to it.
