Alignment Research Critiquer
Nancy Vigil, Ashish Rai
Alignment Research Critiquer is a tool for early career and independent alignment researchers to have access to high-quality feedback loops
Reviewer's Comments
Reviewer's Comments



Marc Carauleanu
Great project! I like how all claims made by the model are supported by evidence from the paper. I would be excited to see the evidence section hidden for convenience and opened when needed. On top of that, it would like to see features being integrated that try to collate evidence from other papers for the claims (both positive and negative).
Jonny Spicer
This is a great project and I’m excited to see that you have detailed next steps to continue. I think it’s reasonably likely that lack of senior mentorship is the largest bottleneck in safety research now, so I think this is trying to solve the most important problem in the domain. Further, I think you’ve made an excellent start. As an additional further step, I’d be excited to see you explore the feasibility of fine-tuning a model. Given the relatively small number of examples usually required for fine-tuning runs, I think by eg collaborating with MATS mentors, you could plausibly get the example data you need.
Jason Schreiber
Working on a good AIS research critique agent seems very promising to me. I’d be excited to see some experimentation on the UX side here and some thoughts on how to collect information on the tool’s performance.
Jacques Thibodeau
I think this project is headed in the right direction! One of the issues with paper critiques from language models is perhaps that the model does not have enough context for what makes a good research paper in a niche within alignment research. Therefore, focusing on specific areas of alignment research (like mechanistic interpretability), we might be able to create significantly better critiques from the models.
Cite this work
@misc {
title={
Alignment Research Critiquer
},
author={
Nancy Vigil, Ashish Rai
},
date={
7/29/24
},
organization={Apart Research},
note={Research submission to the research sprint hosted by Apart.},
howpublished={https://apartresearch.com}
}
Jul 28, 2025
Local Learning Coefficients Predict Developmental Milestones During Group Relative Policy Optimization
In this work, we investigate the emergence of capabilities in reinforcement learning (RL) by framing them as developmental phase transitions. We propose that the individual components of the reward function can serve as direct observables for these transitions, avoiding the need for complex, derived metrics. To test this, we trained a language model on an arithmetic task using Group Relative Policy Optimization (GRPO) and analyzed its learning trajectory with the Local Learning Coefficient (LLC) from Singular Learning Theory. Our findings show a strong qualitative correlation between spikes in the LLC—indicating a phase transition—and significant shifts in the model's behavior, as reflected by changes in specific reward components for correctness and conciseness. This demonstrates a more direct and scalable method for monitoring capability acquisition, offering a valuable proof-of-concept for developmental interpretability and AI safety. To facilitate reproducibility, we make our code available at \url{github.com/ilijalichkovski/apart-physics}.
Read More
Jul 28, 2025
AI agentic system epidemiology
As AI systems scale into decentralized, multi-agent deployments, emergent vulnerabilities challenge our ability to evaluate and manage systemic risks.
In this work, we adapt classical epidemiological modeling (specifically SEIR compartment models) to model adversarial behavior propagation in AI agents.
By solving systems of ODEs describing the systems with physics-informed neural networks (PINNs), we analyze stable and unstable equilibria, bifurcation points, and the effectiveness of interventions.
We estimate parameters from real-world data (e.g., adversarial success rates, detection latency, patching delays) and simulate attack propagation scenarios across 8 sectors (enterprise, retail, trading, development, customer service, academia, medical, and critical infrastructure AI tools).
Our results demonstrate how agent population dynamics interact with architectural and policy design interventions to stabilize the system.
This framework bridges concepts from dynamical systems and cybersecurity to offer a proactive, quantitative toolbox on AI safety.
We argue that epidemic-style monitoring and tools grounded in interpretable, physics-aligned dynamics can serve as early warning systems for cascading AI agentic failures.
Read More
Jul 28, 2025
Momentum–Point-Perplexity Mechanics in Large Language Models
This work analyzes the hidden states of twenty different open-source transformer language models, ranging from small to medium size and covering five major architectures. The key discovery is that these models show signs of "energy conservation" during inference—meaning a certain measure combining changes in hidden states and token unpredictability stays almost constant as the model processes text.
The authors developed a new framework inspired by physics to jointly analyze how hidden states and prediction confidence evolve over time. They propose that transformers' behavior can be understood as following certain mechanical principles, much like how physical systems follow rules like conservation of energy.
Their experiments show that this conserved quantity varies very little between tokens, especially in untrained (random-weight) models, where it's extremely stable. In pre-trained models, the average energy drops more due to training, but there are larger relative fluctuations from token to token.
They also introduce a new method, based on this framework, for controlling transformer outputs by "steering" the hidden states. This method achieves good results—producing completions rated as higher in semantic quality, while still maintaining the same kind of energy stability.
Overall, the findings suggest that viewing transformer models through the lens of physical mechanics gives new, principled ways to interpret and control their behavior. It also highlights a key difference: random models behave more like balanced systems, while trained models make quicker, more decisive state changes at the cost of less precise energy conservation.
Read More
Jul 28, 2025
Local Learning Coefficients Predict Developmental Milestones During Group Relative Policy Optimization
In this work, we investigate the emergence of capabilities in reinforcement learning (RL) by framing them as developmental phase transitions. We propose that the individual components of the reward function can serve as direct observables for these transitions, avoiding the need for complex, derived metrics. To test this, we trained a language model on an arithmetic task using Group Relative Policy Optimization (GRPO) and analyzed its learning trajectory with the Local Learning Coefficient (LLC) from Singular Learning Theory. Our findings show a strong qualitative correlation between spikes in the LLC—indicating a phase transition—and significant shifts in the model's behavior, as reflected by changes in specific reward components for correctness and conciseness. This demonstrates a more direct and scalable method for monitoring capability acquisition, offering a valuable proof-of-concept for developmental interpretability and AI safety. To facilitate reproducibility, we make our code available at \url{github.com/ilijalichkovski/apart-physics}.
Read More
Jul 28, 2025
AI agentic system epidemiology
As AI systems scale into decentralized, multi-agent deployments, emergent vulnerabilities challenge our ability to evaluate and manage systemic risks.
In this work, we adapt classical epidemiological modeling (specifically SEIR compartment models) to model adversarial behavior propagation in AI agents.
By solving systems of ODEs describing the systems with physics-informed neural networks (PINNs), we analyze stable and unstable equilibria, bifurcation points, and the effectiveness of interventions.
We estimate parameters from real-world data (e.g., adversarial success rates, detection latency, patching delays) and simulate attack propagation scenarios across 8 sectors (enterprise, retail, trading, development, customer service, academia, medical, and critical infrastructure AI tools).
Our results demonstrate how agent population dynamics interact with architectural and policy design interventions to stabilize the system.
This framework bridges concepts from dynamical systems and cybersecurity to offer a proactive, quantitative toolbox on AI safety.
We argue that epidemic-style monitoring and tools grounded in interpretable, physics-aligned dynamics can serve as early warning systems for cascading AI agentic failures.
Read More
Jul 28, 2025
Local Learning Coefficients Predict Developmental Milestones During Group Relative Policy Optimization
In this work, we investigate the emergence of capabilities in reinforcement learning (RL) by framing them as developmental phase transitions. We propose that the individual components of the reward function can serve as direct observables for these transitions, avoiding the need for complex, derived metrics. To test this, we trained a language model on an arithmetic task using Group Relative Policy Optimization (GRPO) and analyzed its learning trajectory with the Local Learning Coefficient (LLC) from Singular Learning Theory. Our findings show a strong qualitative correlation between spikes in the LLC—indicating a phase transition—and significant shifts in the model's behavior, as reflected by changes in specific reward components for correctness and conciseness. This demonstrates a more direct and scalable method for monitoring capability acquisition, offering a valuable proof-of-concept for developmental interpretability and AI safety. To facilitate reproducibility, we make our code available at \url{github.com/ilijalichkovski/apart-physics}.
Read More
Jul 28, 2025
AI agentic system epidemiology
As AI systems scale into decentralized, multi-agent deployments, emergent vulnerabilities challenge our ability to evaluate and manage systemic risks.
In this work, we adapt classical epidemiological modeling (specifically SEIR compartment models) to model adversarial behavior propagation in AI agents.
By solving systems of ODEs describing the systems with physics-informed neural networks (PINNs), we analyze stable and unstable equilibria, bifurcation points, and the effectiveness of interventions.
We estimate parameters from real-world data (e.g., adversarial success rates, detection latency, patching delays) and simulate attack propagation scenarios across 8 sectors (enterprise, retail, trading, development, customer service, academia, medical, and critical infrastructure AI tools).
Our results demonstrate how agent population dynamics interact with architectural and policy design interventions to stabilize the system.
This framework bridges concepts from dynamical systems and cybersecurity to offer a proactive, quantitative toolbox on AI safety.
We argue that epidemic-style monitoring and tools grounded in interpretable, physics-aligned dynamics can serve as early warning systems for cascading AI agentic failures.
Read More
Jul 28, 2025
Local Learning Coefficients Predict Developmental Milestones During Group Relative Policy Optimization
In this work, we investigate the emergence of capabilities in reinforcement learning (RL) by framing them as developmental phase transitions. We propose that the individual components of the reward function can serve as direct observables for these transitions, avoiding the need for complex, derived metrics. To test this, we trained a language model on an arithmetic task using Group Relative Policy Optimization (GRPO) and analyzed its learning trajectory with the Local Learning Coefficient (LLC) from Singular Learning Theory. Our findings show a strong qualitative correlation between spikes in the LLC—indicating a phase transition—and significant shifts in the model's behavior, as reflected by changes in specific reward components for correctness and conciseness. This demonstrates a more direct and scalable method for monitoring capability acquisition, offering a valuable proof-of-concept for developmental interpretability and AI safety. To facilitate reproducibility, we make our code available at \url{github.com/ilijalichkovski/apart-physics}.
Read More
Jul 28, 2025
AI agentic system epidemiology
As AI systems scale into decentralized, multi-agent deployments, emergent vulnerabilities challenge our ability to evaluate and manage systemic risks.
In this work, we adapt classical epidemiological modeling (specifically SEIR compartment models) to model adversarial behavior propagation in AI agents.
By solving systems of ODEs describing the systems with physics-informed neural networks (PINNs), we analyze stable and unstable equilibria, bifurcation points, and the effectiveness of interventions.
We estimate parameters from real-world data (e.g., adversarial success rates, detection latency, patching delays) and simulate attack propagation scenarios across 8 sectors (enterprise, retail, trading, development, customer service, academia, medical, and critical infrastructure AI tools).
Our results demonstrate how agent population dynamics interact with architectural and policy design interventions to stabilize the system.
This framework bridges concepts from dynamical systems and cybersecurity to offer a proactive, quantitative toolbox on AI safety.
We argue that epidemic-style monitoring and tools grounded in interpretable, physics-aligned dynamics can serve as early warning systems for cascading AI agentic failures.
Read More

Sign up to stay updated on the
latest news, research, and events

Sign up to stay updated on the
latest news, research, and events

Sign up to stay updated on the
latest news, research, and events

Sign up to stay updated on the
latest news, research, and events