← Back to papers

On Controllability of AI

★ ★ ☆ ☆ ☆

Paper Summary

Paperzilla title
AI: Can't Touch This (Probably)

The paper argues that artificial general intelligence (AGI) and superintelligence (ASI) are fundamentally uncontrollable due to their inherent complexity and potential for self-modification. It suggests that even partial control is unlikely to be achievable in practice and that efforts should focus on mitigating risks associated with uncontrolled AI rather than seeking complete control.

Explain Like I'm Five

Scientists found that super smart robots might be too hard to control, like a really clever pet that can change itself. So, instead of trying to completely boss them around, we should focus on making sure they are safe.

Possible Conflicts of Interest

The author acknowledges partial funding from organizations interested in AI safety, which could potentially bias the research towards emphasizing risks.

Identified Limitations

Focus on Hypothetical Scenarios
The paper focuses on a hypothetical scenario of superintelligent AI, which makes it difficult to evaluate the practical implications of its claims for current AI development.
Lack of Empirical Evidence
The paper primarily relies on philosophical arguments and anecdotal evidence rather than empirical research or formal proofs.
Over-Reliance on Quotations
The paper extensively quotes other authors and researchers but lacks original experimental or quantitative data to support its claims.
Incomplete Literature Review
The paper does not engage with existing literature on AI safety and control in a thorough and balanced way, potentially misrepresenting the current state of the field.
Misunderstanding of Control Theory
The central argument about uncontrollability is based on a simplified and potentially inaccurate understanding of control theory and related disciplines.

Rating Explanation

The paper presents an interesting argument about the potential limits of AI control, but its heavy reliance on philosophical arguments and anecdotal evidence, combined with some inaccuracies and lack of original research, weakens its overall scientific value.

Good to know

This is the Starter analysis. Paperzilla Pro fact-checks every citation, researches author backgrounds and funding sources, and uses advanced AI reasoning for more thorough insights.

Explore Pro →

Topic Hierarchy

File Information

Original Title: On Controllability of AI
Uploaded: July 10, 2025 at 06:29 AM
Privacy: Public