Thousands of AI Authors on the Future of AI
Authors: Katja Grace, Julia Fabienne Sandkühler, Harlan Stewart, Benjamin Weinstein-Raun, Stephen Thomas, Zach Stein-Perlman, John Salvatier, Jan Brauner, Richard C. Korzekwa
JAIR 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | In October 2023, 2,778 researchers who had published in top-tier artificial intelligence (AI) venues gave predictions on the pace, nature and impacts of AI progress. Significant steps were taken to minimize and evaluate bias. In evaluations of participation bias, we found that most groups responded at similar rates. The participants estimated that several milestones had at least a 50% chance of being feasible for AI by 2028, including constructing a payment processing site and fine-tuning an LLM. We conducted two large surveys in 2022 and 2023 of AI researchers who had published peer-reviewed research in the prior year in top AI venues. |
| Researcher Affiliation | Collaboration | KATJA GRACE , AI Impacts, Berkeley, USA JULIA FABIENNE SANDKÜHLER , Department of Psychology, University of Bonn, Germany HARLAN STEWART , AI Impacts, Berkeley, USA BENJAMIN WEINSTEIN-RAUN , Independent, USA STEPHEN THOMAS, AI Impacts, Berkeley, USA ZACH STEIN-PERLMAN, AI Impacts, Berkeley, USA JOHN SALVATIER, Independent, USA JAN BRAUNER , Department of Computer Science, University of Oxford, UK RICHARD C. KORZEKWA , AI Impacts, Berkeley, USA |
| Pseudocode | No | No pseudocode or algorithm blocks are present. The paper describes survey methodology and statistical analysis methods in prose, without structured pseudo-code or algorithm blocks. |
| Open Source Code | No | The anonymized data and full set of questions is available at osf.io/vmdny. This provides access to the survey data and questions, not the source code for the methodology or analysis. The paper does not explicitly state that the source code for their analysis is available. |
| Open Datasets | Yes | The anonymized data and full set of questions is available at osf.io/vmdny. |
| Dataset Splits | No | No traditional dataset splits (e.g., train/test/validation) are mentioned, as the paper describes a survey and its analysis, not the training of a machine learning model on a dataset. The paper discusses participant allocation to different question framings and subsets for survey design, which is not equivalent to dataset splitting for model reproduction. |
| Hardware Specification | No | No specific hardware details are provided, as the paper focuses on survey methodology and statistical analysis rather than computational experiments requiring particular hardware specifications like GPU or CPU models. |
| Software Dependencies | Yes | Data cleaning, analysis, and figure-creation was performed using R statistical software, version 4.3.1, SPSS (Version 29), Google Sheets and Creately. |
| Experiment Setup | No | The paper describes a survey methodology and its analysis, not an experiment involving machine learning model training. Therefore, details like hyperparameters, model initialization, or training schedules are not applicable or provided. |