Society / Social Change
Track social change, shifting values, public sentiment and cultural transformation through structured summaries built from curated sources.
"Collective intelligence requires a broader reimagining of technology and democracy" #TEDTalks
Topic
Collective Intelligence and AI Ethics
Key insights
- Almost three-quarters of people agree that AI should protect free speech
- % of respondents believe AI should not be racist or sexist
- Only around 50% think AI should be funny, indicating contentious issues remain
- The collective constitution project with Anthropic involved retraining a powerful language model on principles from a thousand representative Americans
- The peoples model, based on the co-written constitution, was found to be as capable and fairer than the researcher-developed model
- Participants with little AI experience outperformed full-time researchers in creating a fairer chatbot
Perspectives
short
Support for AI Ethics and Collective Intelligence
- Highlights that almost three-quarters of people agree AI should protect free speech
- Claims 90% agree AI should not be racist or sexist
- Notes that a model trained on a collective constitution was fairer than traditional models
- Argues that people with little AI experience outperformed experts in creating a fair chatbot
- Proposes co-ownership models for data used in AI development
- Emphasizes the need for broader reimagining of technology and democracy
Critique of Traditional AI Development
- Questions the effectiveness of traditional AI models developed by researchers
- Rejects the notion that only experts can create fair AI systems
Neutral / Shared
- Mentions contentious issues regarding AIs role in humor
Metrics
agreement_on_free_speech
Almost three-quarters %
percentage of people who agree AI should protect free speech
Indicates strong public support for free speech principles in AI development.
Almost three-quarters of people agree that AI should protect free speech.
agreement_on_non_discrimination
90%
percentage of respondents who believe AI should not be racist or sexist
Highlights a strong consensus against bias in AI systems.
90% agree that AI should not be racist or sexist.
agreement_on_humor
around 50%
percentage of people who think AI should be funny
Reflects ongoing debates about the role of humor in AI interactions.
Only around 50% think that AI should be funny though, so there's still contentious issues out there.
model_comparison_fairness
more fair
comparison of fairness between people's model and researchers' model
Suggests that non-experts can contribute valuable perspectives in AI development.
the people's model, trained on the co-written constitution, was just as capable and more fair than the model that the researchers had come up with.
performance_of_non_experts
better than researchers
performance comparison between participants with little AI experience and full-time researchers
Indicates that diverse backgrounds can enhance AI development outcomes.
People with little to no experience in AI did better than researchers who work on this full time in building a fairer chatbot.
Key entities
Timeline highlights
00:00–05:00
A significant majority of people believe AI should protect free speech and not exhibit racism or sexism. The collective constitution project with Anthropic demonstrated that a model trained on principles from everyday Americans was fairer and equally capable compared to traditional models developed by researchers.
- Almost three-quarters of people agree that AI should protect free speech
- % of respondents believe AI should not be racist or sexist
- Only around 50% think AI should be funny, indicating contentious issues remain
- The collective constitution project with Anthropic involved retraining a powerful language model on principles from a thousand representative Americans
- The peoples model, based on the co-written constitution, was found to be as capable and fairer than the researcher-developed model
- Participants with little AI experience outperformed full-time researchers in creating a fairer chatbot