I’m a mathematics for quantum physics graduate student considering redirecting my focus toward AI alignment research. My background includes: - Graduate-level mathematics - Focus on quantum physics - Programming experience with Python - Interest in type theory and formal systems
I’m particularly drawn to MIRI-style approaches and interested in: - Formal verification methods - Decision theory implementation - Logical induction - Mathematical bounds on AI systems
My current program feels too theoretical and disconnected from urgent needs. I’m looking to: - Connect with alignment researchers - Find concrete projects to contribute to - Apply mathematical rigor to safety problems - Work on practical implementations
Regarding timelines: I have significant concerns about rapid capability advances, particularly given recent developments (o3). I’m prioritizing work that could contribute meaningfully in a compressed timeframe.
Looking for guidance on: - Most neglected mathematical approaches to alignment - Collaboration opportunities - Where to start contributing effectively - Balance between theory and implementation
If you’re interested in mathematical bounds in AI systems and you haven’t seen it already check out https://arxiv.org/pdf/quant-ph/9908043 Ultimate Physical Limits to Computation by Seth Loyd and related works. Online I’ve been jokingly saying “Intelligence has a speed of light.” Well, we know intelligence involves computation so there has to be some upper bound at some point. But until we define some notion of Atomic Standard Reasoning Unit of Inferential Distance, we don’t have a good way of talking about how much more efficient a computer like you and me are compared to Claude at natural language generation, for example.
Particularly the open problems post (once you know what AIXI is).
For a balance between theory and implementation, I think Michael Cohen’s work on AIXI-like agents is promising.
Also look into Alex Altair’s selection theorems, John Wentworth’s natural abstractions, Vanessa Kosoy’s infra-Bayesianism (and more generally learning theoretic agenda which I suppose I’m part of), and Abram Demski’s trust tiling.
Hi! I’m Embee but you can call me Max.
I’m a mathematics for quantum physics graduate student considering redirecting my focus toward AI alignment research. My background includes:
- Graduate-level mathematics
- Focus on quantum physics
- Programming experience with Python
- Interest in type theory and formal systems
I’m particularly drawn to MIRI-style approaches and interested in:
- Formal verification methods
- Decision theory implementation
- Logical induction
- Mathematical bounds on AI systems
My current program feels too theoretical and disconnected from urgent needs. I’m looking to:
- Connect with alignment researchers
- Find concrete projects to contribute to
- Apply mathematical rigor to safety problems
- Work on practical implementations
Regarding timelines: I have significant concerns about rapid capability advances, particularly given recent developments (o3). I’m prioritizing work that could contribute meaningfully in a compressed timeframe.
Looking for guidance on:
- Most neglected mathematical approaches to alignment
- Collaboration opportunities
- Where to start contributing effectively
- Balance between theory and implementation
If you’re interested in mathematical bounds in AI systems and you haven’t seen it already check out https://arxiv.org/pdf/quant-ph/9908043 Ultimate Physical Limits to Computation by Seth Loyd and related works. Online I’ve been jokingly saying “Intelligence has a speed of light.” Well, we know intelligence involves computation so there has to be some upper bound at some point. But until we define some notion of Atomic Standard Reasoning Unit of Inferential Distance, we don’t have a good way of talking about how much more efficient a computer like you and me are compared to Claude at natural language generation, for example.
Check out my research program:
https://www.lesswrong.com/s/sLqCreBi2EXNME57o
Particularly the open problems post (once you know what AIXI is).
For a balance between theory and implementation, I think Michael Cohen’s work on AIXI-like agents is promising.
Also look into Alex Altair’s selection theorems, John Wentworth’s natural abstractions, Vanessa Kosoy’s infra-Bayesianism (and more generally learning theoretic agenda which I suppose I’m part of), and Abram Demski’s trust tiling.
If you want to connect with alignment researchers you could attend the agent foundations conference at CMU, apply by tomorrow: https://www.lesswrong.com/posts/cuf4oMFHEQNKMXRvr/agent-foundations-2025-at-cmu