System 1

D. Weininger

Zhong-Zhi Li, Duzhen Zhang, Ming-Liang Zhang, Jiaxin Zhang, Zengyan Liu, Yuxuan Yao, Haotian Xu, Junhao Zheng, Pei-Jie Wang, Xiuyi Chen, Yingying Zhang, Fei Yin, Jiahua Dong, Zhijiang Guo, Le Song, Cheng-Lin Liu

Achieving human-level intelligence requires refining the transition from the fast, intuitive System 1 to the slower, more deliberate System 2 reasoning. While System 1 excels in quick, heuristic decisions, System 2 relies on logical reasoning for more accurate judgments and reduced biases. Foundational Large Language Models (LLMs) excel at fast decision-making but lack the depth for complex reasoning, as they have not yet fully embraced the step-by-step analysis characteristic of true System 2 thinking. Recently, reasoning LLMs like OpenAI’s o1/o3 and DeepSeek’s R1 have demonstrated expert-level performance in fields such as mathematics and coding, closely mimicking the deliberate reasoning of System 2 and showcasing human-like cognitive abilities. This survey begins with a brief overview of the progress in foundational LLMs and the early development of System 2 technologies, exploring how their combination has paved the way for reasoning LLMs. Next, we discuss how to construct reasoning LLMs, trace the evolution of various reasoning models, and examine the core methods that enable advanced reasoning behind them. Additionally, we provide an overview of reasoning benchmarks, offering an in-depth comparison of the performance of representative reasoning LLMs. Finally, we explore promising directions for advancing reasoning LLMs and maintain a real-time GitHub Repository to track the latest developments. We hope this survey will serve as a valuable resource to inspire innovation and drive progress in this rapidly evolving field.

Ping Yu, Jing Xu, J. Weston, Ilia Kulikov

Large language models (LLMs) can spend extra compute during inference to generate intermediate thoughts, which helps to produce better final responses. Since Chain-of-Thought (Wei et al., 2022), many such System 2 techniques have been proposed such as Rephrase and Respond (Deng et al., 2023a), System 2 Attention (Weston and Sukhbaatar, 2023) and Branch-Solve-Merge (Saha et al., 2023). In this work we investigate self-supervised methods to ``compile'' (distill) higher quality outputs from System 2 techniques back into LLM generations without intermediate reasoning token sequences, as this reasoning has been distilled into System 1. We show that several such techniques can be successfully distilled, resulting in improved results compared to the original System 1 performance, and with less inference cost than System 2. We posit that such System 2 distillation will be an important feature of future continually learning AI systems, enabling them to focus System 2 capabilities on the reasoning tasks that they cannot yet do well.

J. Webb, A. Chuang, D. Meier, R. V. von Bardeleben, S. Kodali, Robert L. Smith, J. Hausleiter, G. Ong, R. Boone, T. Ruf, I. George, Molly A. Szerlip, M. Näbauer, F. Ali, R. Moss, F. Kreidel, V. Bapat, K. Schnitzler, J. Ye, M. Wild, M. Akodad, D. Deva, A. Chatfield, M. Mack, P. Grayburn, Mark D. Peterson, R. Makkar, Martin B. Leon, R. Hahn, N. Fam

OBJECTIVES The aim of this study was to report the midterm outcomes at 1 year in the expanded first-in-human experience with the transfemoral EVOQUE system (Edwards Lifesciences) for tricuspid regurgitation (TR). BACKGROUND Untreated TR is associated with excess mortality and morbidity. The first-in-human experience with the EVOQUE tricuspid valve replacement system reported favorable 30-day outcomes with no mortality in a compassionate use population. METHODS Twenty-seven patients with severe TR were treated with the EVOQUE system in a compassionate use experience at 7 centers between May 2019 and July 2020. All patients had clinical right-sided heart failure (HF) and were deemed inoperable and unsuitable for transcatheter edge-to-edge repair by the institutional heart teams. The clinical outcomes collected included all-cause mortality, symptom status, TR severity, HF hospitalization, and major adverse cardiovascular events. RESULTS At baseline, all patients (age: 77 ± 8 years, 89% female) were at high surgical risk (mean Society of Thoracic Surgeons score: 8.6% ± 5.5%), with 89% New York Heart Association functional class III/IV. TR was predominantly functional in etiology (19/27, 70%). At 1 year, mortality was 7% (2/27), 70% of patients were New York Heart Association functional class I/II, and 96% and 87% of patients had a TR grade ≤2+ and ≤1+, respectively. Between 30 days and 1 year, 2 patients experienced HF hospitalizations, and 1 patient required a new pacemaker implantation. CONCLUSIONS In this early, compassionate use experience, the transfemoral transcatheter EVOQUE tricuspid valve replacement system demonstrated durable efficacy, persistent improvement in symptom status, and low rates of mortality and HF hospitalizations at a 1-year follow-up. Further studies are underway to validate its efficacy.

M. Kamruzzaman, Gene Louis Kim

Dual process theory posits that human cognition arises via two systems. System 1, which is a quick, emotional, and intuitive process, which is subject to cognitive biases, and System 2, is a slow, onerous, and deliberate process. Prior research in LLMs found that using chain-of-thought (CoT) prompting in LLMs, which has been often compared to System 2 reasoning, can lead to reduced gender bias. Along these lines, we investigate the relationship between bias, CoT prompting, a direct debiasing, and dual process theory modeling in LLMs. We compare zero-shot CoT, debiasing, and dual process theory-based prompting strategies on two bias datasets spanning nine different social bias categories. We incorporate human and machine personas to determine whether LLM modeling of the effects of dual process theory exist independent of explicit persona models or are tied to the LLM's modeling of human-like generation. We find that a human persona, debiasing, System 2, and CoT prompting all tend to reduce social biases in LLMs, though the best combination of features depends on the exact model and bias category -- resulting in up to a 33 percent drop in stereotypical judgments by an LLM.