Extending Human Intelligence Through AI

2026-05-27 · Source: Microsoft Research · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Emerging Technologies & Innovation · Depth: Advanced, medium

Summary

A new interdisciplinary perspective, detailed in the paper "The Origins of Artificial Intelligence in Natural Intelligence," posits that modern AI systems extend structures rooted in human cognition and language, rather than replicating human intelligence. This framework, drawing on Husserlian phenomenology, explains AI's remarkable fluency in tasks like essay writing and code generation, as well as its recurring limitations, such as hallucinations, compositional reasoning gaps, and brittle multimodal reasoning. The research argues that AI systems learn statistical patterns within language, which inherently contains human understanding, but lack the world-directed engagement that anchors meaning. Consequently, AI safety is reframed as a system-level challenge, emphasizing human responsibility, engineering safeguards, and governance ("harnesses") to ensure trustworthy deployment, rather than focusing on "rogue superintelligence" narratives.

Key takeaway

For Directors of AI/ML deploying new systems, recognize that AI extends human cognition, not replaces it. This means your focus must shift from anticipating "rogue AI" to implementing robust system-level safeguards and governance. You should prioritize layered controls, evaluation, and operational oversight, ensuring human builders retain full responsibility for AI behavior. This approach grounds AI deployment in auditable, governable practices, mitigating risks from ungrounded outputs or automated flawed decisions.

Key insights

AI extends human cognition by modeling language's embedded structures, explaining both its power and limits.

Principles

• AI systems extend, not replicate, human intelligence.
• Language contains "sedimented structures" of human understanding.
• AI safety is a system-level challenge, not model-level.

In practice

• Implement layered safeguards and "harnesses" for AI.
• Prioritize governance and auditable AI deployments.
• Focus on human responsibility in AI system design.

Topics

• AI Philosophy
• Human-AI Interaction
• AI Safety
• Large Language Models
• Cognitive Science
• Trustworthy AI

Best for: Research Scientist, AI Scientist, AI Ethicist, Director of AI/ML

Related on AIssential

• Increasing AI Strategic Competence as a Safety Approach — Strategically competent AIs might advocate for an AI pause, offering a new path for managing advanced AI risks.
• AI is becoming a second brain at the expense of your first one — Over-reliance on AI can erode human judgment, leading to disempowerment and the adoption of AI's biases.
• Autonomous AI needs safeguards beyond model-level guardrails, study finds — Model-level safeguards are insufficient for autonomous AI agents, necessitating formal verification for real-world deployment.
• Human layer of AI: How to build human-centered AI safety to mitigate harm and misuse — Effective AI safety requires proactive risk mapping and empowered safety processes to prevent harm and misuse.
• AI #169: New Knowledge — AI is creating new knowledge and utility, but also new risks and regulatory challenges.
• Guiding a Safe Future for AI – Part 1 — AI's automation of intelligence demands collaborative, multi-faceted safety approaches across immediate, societal, catastrophic, and superintelligence risks.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Microsoft Research.