Assistive Intelligence: Alignment & Accountability
From Singularity to Plurality
AUDREY TANG, 2021-10-08. BASED ON “HOW AI FAILS US”, A FORTHCOMING WHITEPAPER IN THE HARVARD SAFRA CENTER DEMOCRACY WHITEPAPER SERIES. WHITEPAPER CONTRIBUTORS: DIVYA SIDDARTH, DARON ACEMOGLU, DANIELLE ALLEN, KATE CRAWFORD, JAMES EVANS, MICHAEL JORDAN, E. GLEN WEYL
A vision of autonomous machine intelligence that aims to achieve and surpass a generality ascribed to human intelligence. Applying this vision implies centralizing decision authority.
A vision of Digital Plurality have already powerfully transformed how we live. Pluralist approaches could rein in AI’s worst excesses.
“To ensure that artificial general intelligence (AGI)—by which we mean highly autonomous systems that outperform humans at most economically valuable work—benefits all of humanity.” — OpenAI Mission statement “Solve intelligence, then use that to solve everything else.” — Demis Hassabis, Co-Founder, DeepMind — Goal of founder Yann LeCun, Facebook AI Research
Human Competition + Autonomy = Centralization
“Human-level AI.”
Issues with human competition and narrow technical benchmarks…
Issues with autonomy and the atomized view of intelligence…
Issues with centralized scale and power concentration…
Towards Digital Plurality
Complementarity Participation
Mutualism
1. Mitigating Autonomy
Ethics Governance
FAT*
2. Beyond Human Competition
Augmentation Comprehensive AI services
Data Collaboratives
3. Re-Decentralization
Human-centered Design Distributed Web
Knowledge Commons
We have a choice.
internet of beings shared reality collaborative learning human experience
internet of beings shared reality collaborative learning human experience
internet of beings shared reality collaborative learning human experience
internet of beings shared reality collaborative learning human experience
internet of beings shared reality collaborative learning human experience
internet of beings shared reality collaborative learning human experience
Plurality