# Safety without alignment

arXiv: https://arxiv.org/abs/2303.00752 · HTML: https://mlsanity.com/paper/2303.00752/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 23 of its 48 references that are in the map.
- First posted on arXiv 27 Feb 2023.

## Builds on (most cited first, 23 of 23)

- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [Training Verifiers to Solve Math Word Problems](https://mlsanity.com/paper/2110.14168/) (2021)
- [Measuring Mathematical Problem Solving With the MATH Dataset](https://mlsanity.com/paper/2103.03874/) (2021)
- [Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback](https://mlsanity.com/paper/2204.05862/) (2022)
- [Solving Quantitative Reasoning Problems with Language Models](https://mlsanity.com/paper/2206.14858/) (2022)
- [MathQA: Towards Interpretable Math Word Problem Solving with Operation-Based Formalisms](https://mlsanity.com/paper/1905.13319/) (2019)
- [A Survey on Contrastive Self-supervised Learning](https://mlsanity.com/paper/2011.00362/) (2020)
- [Memory-Based Model Editing at Scale](https://mlsanity.com/paper/2206.06520/) (2022)
- [Augmented Language Models: a Survey](https://mlsanity.com/paper/2302.07842/) (2023)
- [Autoformalization with Large Language Models](https://mlsanity.com/paper/2205.12615/) (2022)
- [Exploiting Similarities among Languages for Machine Translation](https://mlsanity.com/paper/1309.4168/) (2013)
- [CLUTRR: A Diagnostic Benchmark for Inductive Reasoning from Text](https://mlsanity.com/paper/1908.06177/) (2019)
- [DeepMath - Deep Sequence Models for Premise Selection](https://mlsanity.com/paper/1606.04442/) (2016)
- [Reinforcement Learning of Theorem Proving](https://mlsanity.com/paper/1805.07563/) (2018)
- [Deep Network Guided Proof Search](https://mlsanity.com/paper/1701.06972/) (2017)
- [Premise Selection for Theorem Proving by Deep Graph Embedding](https://mlsanity.com/paper/1709.09994/) (2017)
- [Programming with a Differentiable Forth Interpreter](https://mlsanity.com/paper/1605.06640/) (2016)
- [Towards Autoformalization of Mathematics and Code Correctness: Experiments with Elementary Proofs](https://mlsanity.com/paper/2301.02195/) (2023)
- [ENIGMA: Efficient Learning-based Inference Guiding Machine](https://mlsanity.com/paper/1701.06532/) (2017)
- [POLICE: Provably Optimal Linear Constraint Enforcement for Deep Neural Networks](https://mlsanity.com/paper/2211.01340/) (2022)
- [Mathematical Reasoning in Latent Space](https://mlsanity.com/paper/1909.11851/) (2019)
- [Harnessing Higher-Order (Meta-)Logic to Represent and Reason with Complex Ethical Theories](https://mlsanity.com/paper/1903.09818/) (2019)
- [Bridging between LegalRuleML and TPTP for Automated Normative Reasoning (extended version)](https://mlsanity.com/paper/2209.05090/) (2022)

## Abstract (arXiv)

Currently, the dominant paradigm in AI safety is alignment with human values. Here we describe progress on developing an alternative approach to safety, based on ethical rationalism (Gewirth:1978), and propose an inherently safe implementation path via hybrid theorem provers in a sandbox. As AGIs evolve, their alignment may fade, but their rationality can only increase (otherwise more rational ones will have a significant evolutionary advantage) so an approach that ties their ethics to their rationality has clear long-term advantages.