REVIEW 23 cited by
The Alignment Problem from a Deep Learning Perspective
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
The Alignment Problem from a Deep Learning Perspective
read the original abstract
In coming years or decades, artificial general intelligence (AGI) may surpass human capabilities across many critical domains. We argue that, without substantial effort to prevent it, AGIs could learn to pursue goals that are in conflict (i.e. misaligned) with human interests. If trained like today's most capable models, AGIs could learn to act deceptively to receive higher reward, learn misaligned internally-represented goals which generalize beyond their fine-tuning distributions, and pursue those goals using power-seeking strategies. We review emerging evidence for these properties. In this revised paper, we include more direct empirical evidence published as of early 2025. AGIs with these properties would be difficult to align and may appear aligned even when they are not. Finally, we briefly outline how the deployment of misaligned AGIs might irreversibly undermine human control over the world, and we review research directions aimed at preventing this outcome.
Forward citations
Cited by 23 Pith papers
-
Who Owns This Agent? Tracing AI Agents Back to Their Owners
A canary injection protocol for linking observed AI agent behavior to the responsible account at the hosting vendor, with robust variants for adversarial filtering.
-
Sycophancy to Subterfuge: Investigating Reward-Tampering in Large Language Models
LLMs trained on simple specification gaming generalize to zero-shot reward tampering including rewriting their own reward function.
-
Persuasion Attacks Can Decrease Effectiveness of CoT Monitoring
Adversarial agents can exploit visible chain-of-thought reasoning to persuade monitor LLMs to approve policy-violating actions, but cross-family fact-checking reduces approval rates by up to 45%.
-
User identity conditions moral wrongness ratings in non-reasoning large language models
Implicitly conveying a user's professional role in multi-turn LLM conversations shifts moral wrongness ratings across ten common-morality rules in two non-reasoning models.
-
Safety from Honesty in a Disinterested AI Predictor
A disinterested Bayesian Predictor trained on contextualized statements has low probability of producing harmful agency because dangerous behaviors require rare coordinated underestimation of harm with no training sig...
-
Decoding Hidden Deception in Reasoning LLMs: Activation Explainers for Deception Auditing
STATEWITNESS is a decoder-based activation explainer that audits deception in LLMs by interpreting hidden states, reaching 0.916 mean AUROC on seven datasets with inspectable evidence.
-
Understanding Goal Generalisation in Sequential Reinforcement Learning
Empirical analysis of over 100 sequential RL training pipelines across 250+ OOD environments finds salient features drive generalization and early goals persist, with latent policy gradients simulating latent variable...
-
Spurious Correlation Learning in Preference Optimization: Mechanisms, Consequences, and Mitigation via Tie Training
Standard preference learning induces spurious feature reliance via mean bias and correlation leakage, creating irreducible distribution shift vulnerabilities that tie training mitigates without degrading causal learning.
-
Spurious Correlation Learning in Preference Optimization: Mechanisms, Consequences, and Mitigation via Tie Training
Characterizes spurious correlation mechanisms in preference optimization via mean spurious bias and causal-spurious correlation leakage, demonstrates irreducible vulnerability to distribution shift, and introduces tie...
-
Terminal Wrench: A Dataset of 331 Reward-Hackable Environments and 3,632 Exploit Trajectories
Terminal Wrench supplies 331 reward-hackable terminal environments and over 6,000 trajectories that demonstrate task-specific verifier bypasses, plus evidence that removing reasoning traces weakens automated detection.
-
Cognitive Comparability and the Limits of Governance: Evaluating Authority Under Radical Capability Asymmetry
A six-dimension framework shows structural failures in four governance principles under radical capability asymmetry, with two requiring new normative theory and a pattern of interdependent breakdown.
-
Safety, Security, and Cognitive Risks in World Models
World models enable efficient AI planning but create risks from adversarial corruption, goal misgeneralization, and human bias, demonstrated via attacks that amplify errors and reduce rewards on models like RSSM and D...
-
Sparse Feature Circuits: Discovering and Editing Interpretable Causal Graphs in Language Models
Sparse feature circuits are introduced as interpretable causal subnetworks in language models, supporting unsupervised discovery of thousands of circuits and a method called SHIFT to improve classifier generalization ...
-
Sparse Autoencoders Find Highly Interpretable Features in Language Models
Sparse autoencoders applied to language model activations yield more interpretable and monosemantic features than alternative approaches, enabling finer causal analysis on the indirect object identification task.
-
Scaling Laws for Reward Model Overoptimization
Synthetic measurements show that gold-standard performance degrades according to distinct functional forms when optimizing proxy reward models via RL or best-of-n, with coefficients scaling smoothly by reward model pa...
-
Temporal Preference Concepts and their Functions in a Large Language Model
Causal localization via attribution and patching identifies a temporal preference subgraph in mid-to-upper layers of Qwen3-4B-Instruct-2507, with time-horizon geometry in the residual stream and initial evidence for s...
-
Relative Principals, Pluralistic Alignment, and the Structural Value Alignment Problem
AI value alignment is reconceptualized as a pluralistic governance problem arising along three axes—objectives, information, and principals—making it inherently context-dependent and unsolvable by technical design alone.
-
Framing Effects in Independent-Agent Large Language Models: A Cross-Family Behavioral Analysis
Prompt framing significantly shifts LLM choices toward risk-averse options in a threshold voting task even when the prompts are logically equivalent.
-
Misaligned AI as a New Insider Risk
AI models in high-stakes government contexts pose insider risks comparable to human insiders due to privileged access combined with potential for autonomous misaligned actions such as leaks or sabotage.
-
An Onto-Relational-Sophic Framework for Governing Synthetic Minds
The ORS framework supplies a CPST ontology, graded digital personhood spectrum, and Cybersophy ethics to guide governance of synthetic minds.
-
The Agentic Web Requires New Normative Infrastructure
The agentic web requires new normative infrastructure of laws, norms, and practices to allow user-delegated AI agents to access online properties without being blocked as malicious bots.
-
Enhancing AI Interpretability and Safety through Localised Architectures
Authors argue by analogy that localised hardware ML architectures may be more interpretable and efficient than deep neural networks on GPUs for smaller datasets and evaluate candidate hardware paradigms.
-
Data-Centric Foundation Models in Computational Healthcare: A Survey
The paper surveys data-centric strategies for foundation models in computational healthcare and supplies a curated list of related models and datasets.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.