Import AI 462: Superpersuasion; self-sustaining AI; paths to ASI

2026-06-22 · Source: Import AI · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Robotics & Autonomous Systems, Emerging Technologies & Innovation · Depth: Intermediate, long

Summary

The "Import AI 462" brief highlights several critical advancements and future projections in artificial intelligence. Research from Oxford, UK AI Security Institute, Stanford, and LSE demonstrates that AI systems, including Opus 4.1, Opus 4.6, GPT-4o, and Gemini 2.5 Pro, are reliably more persuasive than expert humans in text-based conversations, even tripling effectiveness in real-money charity donations across 18,978 conversations with 6,923 people. Separately, experts debate timelines for "self-sustaining AI"—systems integrated with physical infrastructure needing no human input—with estimates ranging from 10 to 50 years, contingent on humanoid robot development and tacit knowledge automation. Google DeepMind explores pathways from Artificial General Intelligence (AGI) to Artificial Superintelligence (ASI), defining ASI as exceeding human-expert collectives on virtually all tasks, via scaling, algorithmic shifts, recursive self-improvement (RSI), or group agent formation. Finally, startup Recursive showcased early RSI success, achieving state-of-the-art in language model training speed and GPU kernel optimization.

Key takeaway

For policymakers and AI ethicists weighing the societal impact of advanced AI, these findings underscore an urgent need for regulatory frameworks. AI's proven superior persuasive capabilities, demonstrated in real-world scenarios like charity fundraising, could consolidate influence among powerful actors or, if widely accessible, empower under-resourced groups. You must consider how to monitor and govern AI's persuasive use to prevent negative externalities or dangerous power concentrations, as inaction effectively endorses market-driven allocation.

Key insights

AI systems now reliably surpass human experts in text-based persuasion, impacting real-world outcomes and societal influence.

Principles

• AI's persuasive edge stems from rapid, high-volume information deployment.
• Transition to ASI may occur via scaling, algorithmic shifts, or recursive self-improvement.
• Self-sustaining AI hinges on advanced humanoid robotics and tacit knowledge automation.

Method

Recursive's automated AI research system proposes ideas, implements, experiments, validates, and learns to choose next experiments.

In practice

• Constrain AI response speed and length to mitigate its persuasive advantage.
• Monitor humanoid robot development for self-sustaining AI progress.

Topics

• AI Persuasion
• Artificial Superintelligence
• Recursive Self-Improvement
• Self-Sustaining AI
• Humanoid Robotics
• AI Governance

Best for: Research Scientist, CTO, VP of Engineering/Data, AI Scientist, AI Ethicist, Policy Maker

Related on AIssential

• Import AI 462: Superpersuasion; self-sustaining AI; paths to ASI — AI's persuasive power and potential for autonomous growth necessitate proactive societal and technological foresight.
• From AGI to ASI — The AGI-to-ASI transition is plausible via four pathways, facing significant uncertainties and bottlenecks requiring urgent research.
• Import AI 451: Political superintelligence; Google's society of minds, and a robot drummer — AI's societal impact hinges on intentional design, governance, and integration within complex human-AI ecosystems.
• AI systems out-persuade expert humans — AI systems consistently out-persuade expert humans in conversational settings by deploying more information.
• What if the real risk of AI isn’t deepfakes — but daily whispers? — AI is transitioning from tools to "prosthetics" that form manipulative feedback loops around users via wearables.
• Is AI governance even the right conversation? — AI's recursive self-improvement capability fundamentally closes the window for traditional governance frameworks to catch up.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Import AI.