System 2 Reasoning Capabilities Are Nigh

TL;DR

This paper reviews the progress and remaining challenges in developing neural models capable of human-like System 2 reasoning, suggesting that minimal further advances are needed to achieve this goal.

Contribution

It provides a comprehensive review of current reasoning models and outlines the remaining steps to develop neural systems with human-like System 2 reasoning capabilities.

Findings

Current models are close to achieving System 2 reasoning

Very little additional progress is needed to attain human-like reasoning

The review identifies key remaining steps for future research

Abstract

In recent years, machine learning models have made strides towards human-like reasoning capabilities from several directions. In this work, we review the current state of the literature and describe the remaining steps to achieve a neural model which can perform System~2 reasoning analogous to a human. We argue that if current models are insufficient to be classed as performing reasoning, there remains very little additional progress needed to attain that goal.