OpenAI launches PaperBench to test AI research replication

EditorLuke Juricic
Published 04/02/2025, 01:21 PM
© Reuters.

Investing.com -- On Wednesday, OpenAI announced the release of PaperBench, a new benchmark designed to evaluate the capabilities of AI agents in replicating cutting-edge AI research. This tool is part of OpenAI’s Preparedness Framework, which aims to assess the readiness of AI systems for complex tasks.

PaperBench requires AI agents to accurately replicate 20 significant papers from the International Conference on Machine Learning (ICML) 2024, involving tasks such as comprehending the research, coding, and conducting experiments. Each paper’s replication process is broken down into 8,316 specific tasks, which are assessed using detailed rubrics created in collaboration with the original authors to ensure precision and realism.

The benchmark introduces a novel way to measure AI performance by decomposing the replication of each ICML 2024 Spotlight and Oral paper into smaller, clearly defined sub-tasks. These tasks are then graded based on a set of criteria outlined in the rubrics. To manage the large volume of evaluations, an AI based on a Large Language Model (LLM) has been developed to serve as a judge, automatically grading the AI agents’ attempts to replicate the research.

During the evaluation of several leading AI models on PaperBench, the top-performing agent, Claude 3.5 Sonnet (New), equipped with open-source tools, achieved an average replication score of 21.0%. Additionally, OpenAI conducted an experiment where top machine learning PhD candidates attempted a subset of the tasks from PaperBench. The results indicated that current AI models have not yet surpassed human performance in these tasks.

OpenAI has made the code for PaperBench publicly available, encouraging further research into the engineering capabilities of AI agents. The open-source initiative aims to foster advancements in understanding how AI can be effectively used in AI research replication and development.

This article was generated with the support of AI and reviewed by an editor. For more information see our T&C.

Latest comments

Risk Disclosure: Trading in financial instruments and/or cryptocurrencies involves high risks including the risk of losing some, or all, of your investment amount, and may not be suitable for all investors. Prices of cryptocurrencies are extremely volatile and may be affected by external factors such as financial, regulatory or political events. Trading on margin increases the financial risks.
Before deciding to trade in financial instrument or cryptocurrencies you should be fully informed of the risks and costs associated with trading the financial markets, carefully consider your investment objectives, level of experience, and risk appetite, and seek professional advice where needed.
Fusion Media would like to remind you that the data contained in this website is not necessarily real-time nor accurate. The data and prices on the website are not necessarily provided by any market or exchange, but may be provided by market makers, and so prices may not be accurate and may differ from the actual price at any given market, meaning prices are indicative and not appropriate for trading purposes. Fusion Media and any provider of the data contained in this website will not accept liability for any loss or damage as a result of your trading, or your reliance on the information contained within this website.
It is prohibited to use, store, reproduce, display, modify, transmit or distribute the data contained in this website without the explicit prior written permission of Fusion Media and/or the data provider. All intellectual property rights are reserved by the providers and/or the exchange providing the data contained in this website.
Fusion Media may be compensated by the advertisers that appear on the website, based on your interaction with the advertisements or advertisers.
© 2007-2025 - Fusion Media Limited. All Rights Reserved.