Anthropic, the company behind the AI chatbot Claude, conducted an experiment involving around 1,000 Americans to explore the idea of letting ordinary people shape the rules that govern AI behavior. By allowing public input, Anthropic aims to bridge the gap between public opinion and the AI industry. The experiment resulted in a “Collective Constitutional AI” where the public crowdsourced rules for AI governance. While there were differences compared to the original constitution, both versions performed similarly in training the AI model. This experiment raises questions about the future of AI governance and the involvement of the public in shaping AI behavior.
Democratizing AI Governance: An Anthropic Experiment
Anthropic, the company behind the AI chatbot Claude, has introduced a groundbreaking idea: involving ordinary people in shaping the rules that govern AI behavior. By engaging around 1,000 Americans in their experiment, Anthropic explored a method that could define the future of AI governance.
The Gap Between Public Opinion and the AI Industry
As AI systems become more prevalent, a divide has emerged between the visions of tech leaders and the general public. Surveys have shown that the public prefers a slower pace of AI development and often distrusts the leaders of tech companies. At the same time, AI is increasingly being used in critical processes and decision-making systems across sectors like healthcare, education, and law enforcement. This raises the question of who should determine the guidelines that high-powered AI systems adhere to: tech companies, regulators, politicians, or the public?
Anthropic’s Novel Experiment: Collective Constitutional AI
Anthropic’s experiment proposes an alternative approach to AI governance called “Collective Constitutional AI.” Instead of relying solely on internal sources, the company sought to crowdsource AI rules from the public. This experiment builds on Anthropic’s earlier work on constitutional AI, which trains language models using a set of principles to ensure alignment with human values.
How the Study Worked
Anthropic collaborated with the Collective Intelligence Project, Polis, and PureSpectrum to engage approximately 1,000 US adults in drafting a constitution for an AI system. Participants were presented with a series of principles and their agreement was used to create a crowdsourced constitution. This constitution was then fed into Claude, Anthropic’s AI chatbot.
Results and Insights
The public’s constitution showed notable differences from Anthropic’s in-house constitution. While there was about a 50% overlap, the public’s principles emphasized objectivity, impartiality, accessibility, and positive behavior promotion. Training models with the public’s input resulted in similar performance compared to models trained with Anthropic’s constitution.
Public AI Constitutions: A Realistic Proposition?
A publicly-drafted AI constitution represents an opportunity for the broader society to have a say in the digital ethics that guide machine learning. It can help ensure transparency, accountability, and contextual sensitivity in AI systems. However, challenges remain in ensuring diverse perspectives are included, avoiding oversimplification, and striking a balance between global principles and localized nuances.
The Mechanics of Public Participation
Scaling up constitutional AI could involve approaches like direct democracy, where periodic AI referendums allow people to vote on key principles, or representative democracy, where elected “AI representatives” advocate for public interests. However, developers may need to retain some power to avoid bias towards prevailing views. Checks and balances, periodic reviews, and an “AI constitution court” could help ensure fairness and adaptability.
The Potential Rewards
While the challenges are significant, the potential rewards of a democratic, accountable, and people-centric AI system make it worth striving for. Companies can evolve and stay competitive by embracing AI and leveraging the insights gained from experiments like Anthropic’s.
Practical AI Solutions for Your Company
To evolve your company with AI and stay competitive, consider the following steps:
1. Identify Automation Opportunities: Locate key customer interaction points that can benefit from AI.
2. Define KPIs: Ensure your AI endeavors have measurable impacts on business outcomes.
3. Select an AI Solution: Choose tools that align with your needs and provide customization.
4. Implement Gradually: Start with a pilot, gather data, and expand AI usage judiciously.
For AI KPI management advice and continuous insights into leveraging AI, connect with us at hello@itinai.com or follow us on Telegram t.me/itinainews and Twitter @itinaicom.
Spotlight on a Practical AI Solution: AI Sales Bot
Consider using the AI Sales Bot from itinai.com/aisalesbot to automate customer engagement 24/7 and manage interactions across all customer journey stages. Discover how AI can redefine your sales processes and customer engagement by exploring solutions at itinai.com.
List of Useful Links:
- AI Lab in Telegram @aiscrumbot – free consultation
- Democratizing AI governance: an Anthropic experiment
- DailyAI
- Twitter – @itinaicom
